langflow/docs/versioned_docs/version-1.10.0/API-Reference/flow-devops-sdk.mdx

---
title: Flow DevOps Toolkit SDK
slug: /flow-devops-sdk
---

Use the Flow DevOps Toolkit SDK to version, test, and deploy your flows.

Instead of manually exporting, sharing, and importing flow JSON files from the Langflow UI, the Flow DevOps toolkit offers terminal-based workflows for versioning, environment variables, testing, and deployment.

## Prerequisites

- [Install and start Langflow](/get-started-installation)
- Create a [Langflow API key](/api-keys-and-authentication)
- Install the Langflow `lfx` package

   To install the `lfx` package from PyPI, do the following:

   1. Create a virtual environment:
      ```bash
      uv venv VENV_NAME
      ```

   2. Activate the virtual environment.
      ```bash
      source VENV_NAME/bin/activate
      ```
   3. Install the Langflow LFX package in the virtual environment:
      ```bash
      uv pip install lfx
      ```
   4. Run Flow DevOps Toolkit commands in the virtual environment that has LFX installed.

      Alternatively, you can run `uvx lfx` commands, or run LFX from the `src/lfx` directory in a cloned Langflow repo.
      For more information, see the [Langflow LFX README](https://github.com/langflow-ai/langflow/blob/main/src/lfx/README.md).

## Create a project and version a flow

1. Create a flow in the Langflow UI, such as the Simple Agent starter flow in the [Quickstart](/get-started-quickstart).
2. Open a terminal session within the virtual environment that has `lfx` installed.
3. To initialize a project, run:
    ```bash
    lfx init PROJECT_NAME
    ```

    Replace `PROJECT_NAME` with a name for your project folder.
    `lfx init` creates a scaffold for your project.
    The output is similar to the following:

    ```text
    demo-project/
    ├── .github/
    │   └── workflows/
    │       ├── langflow-push.yml       # CI workflow
    │       ├── langflow-test.yml       # CI workflow
    │       └── langflow-validate.yml   # CI workflow
    ├── .gitignore                      # ignores legacy credentials file
    ├── .lfx/
    │   └── environments.yaml           # edit with your instance URLs + API key env var names (safe to commit)
    ├── ci/
    │   ├── ci-push.sh                  # generic CI script
    │   ├── ci-test.sh                  # generic CI script
    │   └── ci-validate.sh              # generic CI script
    ├── flows/
    │   └── .gitkeep                    # versioned empty directory
    └── tests/
        ├── __init__.py
        └── test_flows.py               # flow_runner example tests

    ✔ Project scaffolded. Next steps:
      1. Edit `.lfx/environments.yaml` with your instance URL
      2. export LANGFLOW_LOCAL_API_KEY=<key>   (Settings -> API Keys)
      3. lfx pull --env local --output-dir flows/
    ```

    The project scaffold includes the following tools for building flows:

    * `.github/workflows`: GitHub CI tooling.
    * `.lfx/environments.yaml`: Control your project's URL and API keys as environment variables for local, staging, and production environments.
    * `ci/` Shell scripts for pushing, testing, and validating flows.
    * `flows/`: An empty directory to store flows that includes a `.gitkeep` file for flow versioning.
    * `tests/test_flows.py`: Example tests that you can modify to test flows.

4. Add your Langflow API key to your `.env` file, or export it within the terminal session.
    The Flow DevOps SDK includes `url` and `api_key_env` environment variables for `local`, `staging`, and `production` environments.
    The variable name for the API key differs between environments, so ensure you're adding the correct variable.
    For example, to add a Langflow API key to a local Langflow server, set:

    ```text
    export LANGFLOW_LOCAL_API_KEY=LANGFLOW_API_KEY
    ```

5. To test server authentication with your API key, run:

    ```bash
    lfx login
    ```

    The Flow DevOps SDK tests your key against the URL and confirms the connection is working.
    If the test reports `Authentication failed`, create and export a new key and try again.

6. To check the connected server for existing flows, run:

    ```bash
    lfx pull
    ```

    The Flow DevOps SDK lists your server's flows. Output is similar to the following (from a project directory such as `demo-project/`):

    ```text
    Pulling all flows from http://localhost:7860
    Pulled 'Simple Agent' -> flows/Simple_Agent.json

    ┌────────────────┬──────────────────────────────────────┬───────────────────────────┬──────────┐
    │ Name           │ ID                                   │ File                      │ Status   │
    ├────────────────┼──────────────────────────────────────┼───────────────────────────┼──────────┤
    │ Simple Agent   │ c2f91b01-9a73-4c62-b7f0-e15bc3bd6802 │ flows/Simple_Agent.json   │ CREATED  │
    └────────────────┴──────────────────────────────────────┴───────────────────────────┴──────────┘

    1 updated.
    ```

    `lfx pull` pulls flow changes on the server to JSON files under `flows/` in your project (for example `demo-project/flows`).
    If you run `lfx pull` again, the Flow DevOps SDK reports a Status of `Unchanged`.
    You will pull changes in the next steps.
7. In the Langflow UI, open the Simple Agent flow and change the flow.
    For example, change the **Chat Input** to a different input string.
    Save the flow.
8. Return to your terminal, and run `lfx status`.
   The Flow DevOps SDK reports the flow's Status as `UPDATED`, because the hash of the flow JSON has changed with your update.
9. To pull the reported changes from the Langflow server to your local project folder, run `lfx pull`.
10. To _push_ flow changes from flows stored locally in `demo-project/flows` to the Langflow server, run `lfx push`.

## Validate flows

The Flow DevOps SDK can validate that local flows are correctly formed before pushing to the Langflow with `lfx validate`.

1. To test the Simple Agent starter flow, pass the flow JSON path to the `lfx validate` command:
    ```bash
    lfx validate flows/Simple_Agent.json
    ```
2. Once validated, push flow changes to the server with `lfx push`.

## Generate requirements.txt for flows

The Flow DevOps SDK can generate a `requirements.txt` file for a flow.

A flow JSON describes nodes and wiring, and does not list the PyPI packages components import at runtime.
Generate a `requirements.txt` file to capture the minimal Python dependencies, so you can install a matching environment for the flow.

1. From your project directory, point `lfx requirements` at a flow JSON file.
   To print the requirements to the terminal:

   ```bash
   lfx requirements flows/Simple_Agent.json
   ```

   To write a `requirements.txt` file instead of printing, use `-o` or `--output`:

   ```bash
   lfx requirements flows/Simple_Agent.json -o requirements.txt
   ```

2. Optionally, you can now share and serve the flow by keeping the flow JSON and `requirements.txt` in the same environment.
   To serve the flow without the Langflow UI, do the following:

   1. Create a virtual environment:
   ```bash
   uv venv VENV_NAME
   ```

   2. Activate the virtual environment.
   ```bash
   source VENV_NAME/bin/activate
   ```

   3. Install the dependencies from `requirements.txt` in the virtual environment:
   ```bash
   uv pip install -r requirements.txt
   ```

   4. To set a Langflow API key, run:
   ```bash
   export LANGFLOW_API_KEY=LANGFLOW_API_KEY
   ```

   5. To serve the flow without the Langflow UI, pass the flow JSON path to the `lfx serve` command:
   ```bash
   lfx serve flows/Simple_Agent.json
   ```

   `lfx serve` starts a FastAPI app that exposes your flow as an HTTP API endpoint.
   For more information, see the [Langflow LFX README](https://github.com/langflow-ai/langflow/blob/main/src/lfx/README.md).

## Manage multiple environments with `environments.yaml`

The `environments.yaml` file created at initialization contains three example entries for deployment environments:

```yaml
local:
  url: http://127.0.0.1:7860
  api_key_env: LANGFLOW_LOCAL_API_KEY
staging:
  url: https://staging.example.com
  api_key_env: LANGFLOW_STAGING_API_KEY
production:
  url: https://langflow.example.com
  api_key_env: LANGFLOW_PRODUCTION_API_KEY
```

Each entry contains a `url` for the Langflow base URL, and an `api_key_env` field.
The `api_key_env` field names an environment variable that you either `export` or store in a `.env` file, and does not store the secret string itself, which makes `environments.yaml` safe to commit to version control.

The names `local`, `staging`, and `production` in `environments.yaml` are conventions, and can be named whatever your project requires. You can add more than three entries.

`environments.yaml` is distinct from the Langflow or LFX `.env` file.
`environments.yaml` controls which remote Langflow instance you're deploying flows to, flow versioning, and environment variable _names_ for API keys.
The `.env` contains runtime values for the Langflow server, and might also contain _actual secret values_, so the `.env` should not be committed to version control.

Commands that call a Langflow server over HTTP, such as `lfx pull` or `lfx push`, use `--env ENVIRONMENT_NAME` to determine which Langflow instance to send the request to.

For example, to send a `push` request to a server named `local` in `environments.yaml`, run:

```bash
lfx push --env local
```

This command will send the request to the Langflow base URL at `http://127.0.0.1:7860` using a Langflow API key named `LANGFLOW_LOCAL_API_KEY`.