Files
langflow/BUNDLE_API.md
Eric Hare 7896b9f7f1 feat(bundles): metapackage split (Phase A) — engine-only lfx, lfx-bundles long tail, 5 graduated partner packages (#13563)
* feat(extension): add manifest-less lfx.bundles discovery + precedence tier

Foundation for the bundle metapackage split (1.11). Adds a third
@official-slot discovery source: a distribution declaring the
[project.entry-points."lfx.bundles"] group ships a package whose immediate
subdirectories are each a manifest-less bundle, folder-walked and registered
at @official with no extension.json (the langchain-community model).

- new loader/_bundles_root.py::load_lfx_bundles_extensions, mirroring
  discover_inline_bundles but sourcing roots from the lfx.bundles entry-point
  group via find_spec and reusing _load_bundle_directory
- discovery precedence becomes installed > seed > lfx_bundles > dev > inline
  so a manifest-shipping lfx-<provider> always shadows the same-named
  provider in the metapackage (lets a provider graduate with no lockstep
  release), emitting the existing bundle-shadowed warning
- new bundle-discovery-malformed warning code, emitted on warnings (never
  flips ok / aborts startup) for unresolvable declarations and invalid
  provider folder names
- exported via loader/__init__ and the lfx.extension PEP-562 lazy surface

10 new tests in test_load_lfx_bundles.py; full extension unit suite (449)
passes; ruff + format clean; mypy clean on the new module.

* docs(bundles): changelog entry for the lfx.bundles discovery surface

The BUNDLE_API.md changelog gate diffs each PR's branch against main, so
the entry covering this PR's surface additions (lfx.bundles discovery,
the lfx_bundles precedence tier, bundle-discovery-malformed) must live
on this branch, not only further up the stack. Text is verbatim from the
graduate-partners commit so the stacked merge dedupes trivially.

* fix(extension): reject plain-module lfx.bundles targets; correct validate wording

CodeRabbit review fixes on #13563:
- _spec_package_dir no longer falls back to a plain module's parent
  directory -- a single-file entry-point target now emits the
  bundle-discovery-malformed warning instead of folder-walking unrelated
  sibling directories as bundles (+ regression test)
- BUNDLE_API.md / loader docstring no longer say manifest-less providers
  are 'exempt from lfx extension validate'; they are not valid validator
  input (validate requires a manifest and reports manifest-not-found)

* fix(extension): review fixes — shadowed metapackage providers never import; typed manifest-less reload refusal

Review findings on #13563:

- A manifest-less provider whose bundle name is already claimed by a
  higher-precedence @official source (installed/seed) is now skipped
  WITHOUT importing: all @official sources share the
  _lfx_ext.official.<bundle>.* sys.modules namespace, so importing the
  losing copy overwrote the winner's live modules even though
  _resolve_bundle_shadowing dropped its components afterwards.  The
  orchestrator passes the claimed-name map (built by the new
  _claimed_official_bundles helper, mirroring the resolver's winner
  rule); the skipped copy still emits the typed bundle-shadowed warning.
  This collision is the expected post-graduation state, not an error.

- Manifest-less records now carry a manifestless provenance flag
  (additive field on LoadResult + BundleRecord).  The reload pipeline
  refuses them with the new typed reload-manifestless-unsupported code
  instead of routing through load_extension and failing with a
  misleading manifest-not-found.  Hot reload of metapackage providers
  is a process-restart concern (pip-installed content); a loader-based
  reload path can follow if QA wants it.

BUNDLE_API.md changelog entry extended for both surface changes.

* fix(extension): harden lfx.bundles discovery per review — broad find_spec guard, namespace portions, per-mode error codes

Review findings from ogabrielluiz on #13563:

- find_spec on a dotted declaration imports the PARENT package, whose
  __init__ can raise anything; a non-Import error escaped the old catch
  tuple and (through the palette cache's catch-all) wiped every source's
  components for the boot.  Catch Exception and degrade to the malformed
  sentinel; comment corrected.

- Namespace packages: walk ALL submodule_search_locations portions (one
  root per portion) instead of only locations[0], and dedupe resolved
  roots by path so duplicate entry-point declarations (or overlapping
  portions) never walk a directory twice and self-shadow.
  _spec_package_dir -> _spec_package_dirs.

- Split bundle-discovery-malformed into one code per failure mode,
  mirroring the inline tier: bundles-provider-name-invalid (rename the
  directory), bundles-root-unreadable (check permissions), keeping
  bundle-discovery-malformed for unresolvable declarations (fix the
  entry-point).  Same-tier duplicates get duplicate-lfx-bundles-provider
  instead of overloading bundle-shadowed, whose rendered template
  (format_extension_error renders templates, not ad-hoc messages) was
  wrong on both counts for that case.

- The claimed-bundles cross-source skip now emits bundle-shadowed on
  errors (matching _resolve_bundle_shadowing) so filtering by code never
  mixes severities; bundle-shadowed is already in the CLI warn-only set.

3 new regression tests (raising parent package, namespace portions,
duplicate entry points); BUNDLE_API changelog updated.

* feat(bundles): create lfx-bundles metapackage skeleton (#13564)

* feat(bundles): add lfx-bundles metapackage skeleton

Creates the manifest-less lfx-bundles distribution (the langchain-community
model) that the long-tail providers will move into. Empty skeleton for now;
the bulk move (scripts/migrate/consolidate_bundles.py) populates the provider
folders + per-provider extras later.

- src/bundles/lfx-bundles: a single pyproject declaring the lfx.bundles
  entry-point (lfx_bundles = "lfx_bundles"), an lfx>=1.10.0,<2.0.0 pin, and a
  generated (currently empty) `all` extra; plus a bare lfx_bundles namespace
  package and a README documenting the model + install stories
- wired into the root workspace via the existing bundle marker blocks:
  dep lfx-bundles[all]>=1.0,<2.0, uv source, and member
- hyphen dir name so release.yml's src/bundles/*/pyproject.toml glob builds
  it with zero workflow change

Verified: the wheel builds (entry_points.txt carries the lfx.bundles group);
load_lfx_bundles_extensions (PR-1) discovers the entry point and the empty
skeleton registers zero providers with no error.

* fix(ci): nightly bundle rename follows [extras] refs and self-refs

Two gaps in update_bundle_versions.py around extras suffixes, both fatal
or silently wrong for the first nightly carrying the lfx-bundles
metapackage:

- update_root_pyproject_for_bundle's dep regex required the version
  specifier immediately after the name, so "lfx-bundles[all]>=1.0,<2.0"
  (a MAIN dep) was left unrewritten while the workspace member was
  renamed to lfx-bundles-nightly; uv lock then tries to resolve stable
  lfx-bundles from PyPI, where it does not exist. The same gap silently
  left "lfx-docling[local]>=0.1.0" optional-dep refs pointing at the
  stable distribution. The regex now tolerates an [extras] group and
  carries it into the replacement.

- rename_bundle_pyproject skipped self-referencing extras, so the
  metapackage's generated `all` extra kept 45 "lfx-bundles[<provider>]"
  members after the rename, pulling the stable distribution (same
  lfx_bundles import package, install collision) once published. Self
  refs now follow the rename, idempotently.

Tests drive the real script module, mirroring test_bundle_lfx_pin.py.
This closes the PR-2 audit item flagged when the metapackage was
introduced. The canonical-pre-release cutover would retire the nightly
rename entirely; until it lands, the rename must be correct.

* fix(bundles): address review — drop orphaned nightly bundle-rename script, fix README install stories

- Delete scripts/ci/update_bundle_versions.py and its tests: the nightly
  bundle-rename track it served was retired by the canonical pre-release
  cutover (src/bundles/NIGHTLY.md) and no workflow or Makefile target on
  main or any release branch invokes it. Stale doc references in
  sync_bundle_lfx_pin.py and test_bundle_lfx_pin.py updated to match.
- README: split the install section into what works today (langflow, bare
  lfx) vs what arrives with the bulk move / engine-only split
  (lfx[bundles], per-provider lfx-bundles[<provider>] extras), and note
  the generated all extra is empty until the first provider tranche lands.

* ci(bundles): cross-bundle test matrix (lfx contract axis) (#13566)

* ci(bundles): add cross-bundle test matrix (lfx contract axis)

Tests every extracted bundle (the lfx-bundles metapackage + each graduated
lfx-<provider>) against the lfx contract surface it depends on -- overdue
since 1.10 left 4 independently-versioned bundles depending on lfx.

- .github/workflows/cross-bundle-test.yml: discovers bundles via the same
  src/bundles/*/pyproject.toml glob release.yml uses, then per
  (bundle x python 3.10/3.13): installs the in-repo lfx + the bundle, imports
  the declared entry-point package, asserts lfx.bundles discovery is
  error-free (for the metapackage), runs `lfx extension validate` for
  manifest bundles, and runs the bundle's own tests/.
- Triggers: pull_request (src/bundles, src/lfx), workflow_dispatch, a weekly
  schedule, and workflow_call.

The lfx-minor axis seeds with the in-repo lfx (the 1.10 line is unpublished);
when minors publish, add the version dimension (oldest+latest get full tests,
every supported minor gets contract-smoke) per the epic's cost-control shape.

Verified: the workflow YAML parses and the contract-smoke logic imports a
real bundle (lfx_arxiv) cleanly.

* fix(ci): cross-bundle smoke tolerates extras-less SDK degradation

Review findings on the cross-bundle matrix:

- CRITICAL: the contract smoke asserted zero discovery errors, but the CI
  venv installs lfx-bundles WITHOUT per-provider extras, so providers whose
  modules import their SDK at top level degrade with module-import-failed --
  the expected graceful-degradation contract. The smoke now fails only on
  structural codes and reports the degraded-module count. (Graduated partner
  bundles carry their deps directly, not as extras, so their steps are
  unaffected.)
- the scheduled run now actually delivers the exhaustive grid the header
  promised (all supported Pythons on schedule; oldest+latest on PRs)
- concurrency group with cancel-in-progress on PRs so stacked bundle PRs
  don't queue N-bundles x N-pythons jobs per push

* ci(bundles): tomli fallback for the py3.10 cross-bundle smoke

tomllib is stdlib only from Python 3.11, so the contract-smoke heredoc
failed with ModuleNotFoundError on every py3.10 matrix leg. Install the
tomli backport into the smoke venv and import it as a fallback.

* fix(test): coherent sys.modules restore in the ibm without_ibm_db fixture

The cross-bundle matrix's py3.10 leg exposed test-ordering pollution in
the ibm bundle suite: without_ibm_db monkeypatch-deleted three lfx_ibm
modules and re-imported them mid-test, but entries created during the
test survive teardown (delitem with raising=False records nothing for
keys it didn't find) and re-imported parents never regain the submodule
attribute bindings that mock.patch target resolution walks on Python
3.10 (3.11+ mock resolves via sys.modules and tolerates the
incoherence). Two fixture uses in sequence left the tree inconsistent
and 29 later watsonx tests failed with AttributeError on the package.

Snapshot, drop, and restore the entire lfx_ibm module tree wholesale
instead. Full suite now passes on both 3.10 and 3.13.

* feat(lfx): add lfx[bundles] extra and keep lfx engine-only (#13565)

* feat(bundles): add lfx-bundles metapackage skeleton

Creates the manifest-less lfx-bundles distribution (the langchain-community
model) that the long-tail providers will move into. Empty skeleton for now;
the bulk move (scripts/migrate/consolidate_bundles.py) populates the provider
folders + per-provider extras later.

- src/bundles/lfx-bundles: a single pyproject declaring the lfx.bundles
  entry-point (lfx_bundles = "lfx_bundles"), an lfx>=1.10.0,<2.0.0 pin, and a
  generated (currently empty) `all` extra; plus a bare lfx_bundles namespace
  package and a README documenting the model + install stories
- wired into the root workspace via the existing bundle marker blocks:
  dep lfx-bundles[all]>=1.0,<2.0, uv source, and member
- hyphen dir name so release.yml's src/bundles/*/pyproject.toml glob builds
  it with zero workflow change

Verified: the wheel builds (entry_points.txt carries the lfx.bundles group);
load_lfx_bundles_extensions (PR-1) discovers the entry point and the empty
skeleton registers zero providers with no error.

* feat(lfx): add lfx[bundles] extra and keep lfx engine-only

lfx ships no bundles by default. The new optional extra pulls the long-tail
metapackage for users who want the provider components without the full
langflow server install.

- src/lfx/pyproject.toml: [project.optional-dependencies]
  bundles = ["lfx-bundles[all]>=1.0,<2.0"]; intentionally NO lfx[all]
- pip install lfx          -> engine only
  pip install "lfx[bundles]" -> engine + the lfx-bundles long tail
  (documented as a headless/serverless deployment footnote, not a headline)

Verified against the built lfx wheel METADATA: lfx-bundles appears only under
`extra == "bundles"`, never in core Requires-Dist, so the engine-only default
is preserved.

* fix(ci): nightly bundle rename follows [extras] refs and self-refs

Two gaps in update_bundle_versions.py around extras suffixes, both fatal
or silently wrong for the first nightly carrying the lfx-bundles
metapackage:

- update_root_pyproject_for_bundle's dep regex required the version
  specifier immediately after the name, so "lfx-bundles[all]>=1.0,<2.0"
  (a MAIN dep) was left unrewritten while the workspace member was
  renamed to lfx-bundles-nightly; uv lock then tries to resolve stable
  lfx-bundles from PyPI, where it does not exist. The same gap silently
  left "lfx-docling[local]>=0.1.0" optional-dep refs pointing at the
  stable distribution. The regex now tolerates an [extras] group and
  carries it into the replacement.

- rename_bundle_pyproject skipped self-referencing extras, so the
  metapackage's generated `all` extra kept 45 "lfx-bundles[<provider>]"
  members after the rename, pulling the stable distribution (same
  lfx_bundles import package, install collision) once published. Self
  refs now follow the rename, idempotently.

Tests drive the real script module, mirroring test_bundle_lfx_pin.py.
This closes the PR-2 audit item flagged when the metapackage was
introduced. The canonical-pre-release cutover would retire the nightly
rename entirely; until it lands, the rename must be correct.

* ci(bundles): freeze lfx/components against new top-level providers (#13567)

* feat(bundles): add lfx-bundles metapackage skeleton

Creates the manifest-less lfx-bundles distribution (the langchain-community
model) that the long-tail providers will move into. Empty skeleton for now;
the bulk move (scripts/migrate/consolidate_bundles.py) populates the provider
folders + per-provider extras later.

- src/bundles/lfx-bundles: a single pyproject declaring the lfx.bundles
  entry-point (lfx_bundles = "lfx_bundles"), an lfx>=1.10.0,<2.0.0 pin, and a
  generated (currently empty) `all` extra; plus a bare lfx_bundles namespace
  package and a README documenting the model + install stories
- wired into the root workspace via the existing bundle marker blocks:
  dep lfx-bundles[all]>=1.0,<2.0, uv source, and member
- hyphen dir name so release.yml's src/bundles/*/pyproject.toml glob builds
  it with zero workflow change

Verified: the wheel builds (entry_points.txt carries the lfx.bundles group);
load_lfx_bundles_extensions (PR-1) discovers the entry point and the empty
skeleton registers zero providers with no error.

* ci(bundles): freeze lfx/components against new top-level providers

After the metapackage split, new providers go to lfx-bundles (or a graduated
lfx-<provider>), never in-tree. Adds an additions-only CI gate.

- scripts/ci/check_components_frozen.py: stdlib gate comparing the live
  top-level dir listing of src/lfx/src/lfx/components/ against a committed
  baseline; fails on any directory not in the baseline. Removals are allowed
  so M4 shim cleanup (which shrinks the set) never trips it.
- scripts/ci/frozen_component_dirs.txt: the 111-dir baseline.
- .github/workflows/lint-py.yml: new freeze-components job runs the gate
  (stdlib-only, no uv sync).

Verified locally: passes on the baseline, fails (exit 1) on a simulated new
provider directory with an actionable message, passes again after cleanup.

* fix(ci): freeze gate counts only dirs shipping __init__.py

Review finding: a stray directory holding only __pycache__ bytecode (left
behind by a branch switch) tripped the additions-only gate locally. A
directory without __init__.py is not a provider; ignore it.

* fix(ci): nightly bundle rename follows [extras] refs and self-refs

Two gaps in update_bundle_versions.py around extras suffixes, both fatal
or silently wrong for the first nightly carrying the lfx-bundles
metapackage:

- update_root_pyproject_for_bundle's dep regex required the version
  specifier immediately after the name, so "lfx-bundles[all]>=1.0,<2.0"
  (a MAIN dep) was left unrewritten while the workspace member was
  renamed to lfx-bundles-nightly; uv lock then tries to resolve stable
  lfx-bundles from PyPI, where it does not exist. The same gap silently
  left "lfx-docling[local]>=0.1.0" optional-dep refs pointing at the
  stable distribution. The regex now tolerates an [extras] group and
  carries it into the replacement.

- rename_bundle_pyproject skipped self-referencing extras, so the
  metapackage's generated `all` extra kept 45 "lfx-bundles[<provider>]"
  members after the rename, pulling the stable distribution (same
  lfx_bundles import package, install collision) once published. Self
  refs now follow the rename, idempotently.

Tests drive the real script module, mirroring test_bundle_lfx_pin.py.
This closes the PR-2 audit item flagged when the metapackage was
introduced. The canonical-pre-release cutover would retire the nightly
rename entirely; until it lands, the rename must be correct.

* feat(bundles): move 45 long-tail providers into lfx-bundles + graduate 5 partner packages (#13568)

* feat(bundles): add lfx-bundles metapackage skeleton

Creates the manifest-less lfx-bundles distribution (the langchain-community
model) that the long-tail providers will move into. Empty skeleton for now;
the bulk move (scripts/migrate/consolidate_bundles.py) populates the provider
folders + per-provider extras later.

- src/bundles/lfx-bundles: a single pyproject declaring the lfx.bundles
  entry-point (lfx_bundles = "lfx_bundles"), an lfx>=1.10.0,<2.0.0 pin, and a
  generated (currently empty) `all` extra; plus a bare lfx_bundles namespace
  package and a README documenting the model + install stories
- wired into the root workspace via the existing bundle marker blocks:
  dep lfx-bundles[all]>=1.0,<2.0, uv source, and member
- hyphen dir name so release.yml's src/bundles/*/pyproject.toml glob builds
  it with zero workflow change

Verified: the wheel builds (entry_points.txt carries the lfx.bundles group);
load_lfx_bundles_extensions (PR-1) discovers the entry point and the empty
skeleton registers zero providers with no error.

* feat(bundles): consolidate first long-tail tranche into lfx-bundles

Adds scripts/migrate/consolidate_bundles.py (the inverse of port_bundle.py --
moves in-tree providers into the manifest-less lfx-bundles metapackage) and
runs it on a verified 5-provider tranche: tavily, exa, wikipedia, yahoosearch,
wolframalpha.

Per provider the script:
- moves src/lfx/src/lfx/components/<p>/ -> lfx_bundles/<p>/ (lowercase names);
- leaves a fail-soft import shim (first line `# lfx-bundles-shim`) so
  `from lfx.components.<p> import X` keeps working when lfx-bundles is
  installed, and raises an actionable ImportError otherwise;
- merges the provider's third-party deps into a PEP 685-normalized lfx-bundles
  extra and regenerates the `all` aggregate. Dep parity holds: `uv sync` is a
  no-op because those deps were already pulled via langflow-base[complete];
- appends the 4-entry migration block per Component class (28 entries) so saved
  flows referencing lfx.components.<p>.<Class> migrate to
  ext:<p>:<Class>@official.

To avoid double registration, the in-tree component walk
(_load_components_dynamically) now skips shimmed provider dirs, and
component_index.json is regenerated (355->348 components, 95->90 modules); the
moved providers load only at @official via lfx.bundles discovery.

Verified: discovery finds all 5 at @official with no errors; shims resolve;
`import lfx.components` still works; the index drops the 7 moved component
entries (residual `tools`-category name refs resolve via the shim); ruff clean.

First tranche proves the engine; the remaining long-tail scales by extending
PROVIDER_DEPS (each provider's deps verified individually -- the careful part).

* feat(bundles): consolidate 30-provider tranche 2 into lfx-bundles

Extends PROVIDER_DEPS with 30 individually-verified providers and runs the
consolidation: vector stores (chroma, clickhouse, couchbase, milvus, mongodb,
pgvector, pinecone, qdrant, supabase, upstash, weaviate), model providers
(groq, mistral, ollama, perplexity, sambanova), and tools/memory/data (apify,
assemblyai, confluence, firecrawl, git, glean, icosacomputing, mem0, needle,
scrapegraph, serpapi, unstructured, youtube, zep).

Dep verification (the careful part): every spec comes from langflow-base's
per-provider extras or its direct dependencies; langchain-community providers
carry the wrapper plus the SDK the wrapper lazy-imports (e.g. pgvector,
atlassian-python-api); requests is declared explicitly where imported (it is
only transitive in today's env); pinecone keeps its python_version<'3.14'
marker verbatim. Tranche excludes: providers with langflow imports (vlmrun),
provider-specific lfx.base dirs (composio/huggingface/langwatch), case-
sensitive names (FAISS/Notion), the openai-SDK family (azure/aiml/deepseek/
litellm/lmstudio/novita/openrouter/vllm/xai/cometapi -- cleaner after PR-8),
and the partner set.

Dep parity verified at the resolution level: the uv.lock diff is +220 lines of
lfx-bundles extras metadata with ZERO packages added or removed (`name =` diff
empty), so pip install langflow resolves the identical set.

Also: 192 append-only migration entries (48 classes x 4, zero bare-name
ambiguities); component_index.json regenerated 348->300 components / 90->60
modules (exactly the moved set, no stale standalone entries); mongodb_atlas.py
SLF001 per-file-ignore and the mem0/mongodb detect-secrets baseline entries
migrated to the new paths (lint/secrets exceptions travel with moved files);
35 bundles now discover at @official with 55 components; shims verified across
categories; 449 extension tests pass.

* feat(bundles): consolidate openai-SDK family tranche 3 into lfx-bundles

10 providers that ride the langchain-openai wrapper, deferred from tranche 2
until the partner graduation settled the openai-SDK dep story: aiml, azure,
cometapi, deepseek, litellm, lmstudio, novita, openrouter, vllm, xai.

Dep verification: every provider declares langchain-openai>=1.1.6; the openai
SDK is declared only where a component imports it directly (aiml, deepseek,
litellm, lmstudio, vllm, xai -- wrapper-transitive elsewhere); requests
declared where imported (cometapi, deepseek, novita, xai); lmstudio's lazy
NVIDIAEmbeddings path gets langchain-nvidia-ai-endpoints~=1.0.0. The litellm
component drives LiteLLM-served endpoints through the OpenAI client and does
NOT import the litellm package -- langflow-base's litellm extra stays put.
These providers use the lazy _dynamic_imports __init__ shape; it survives the
move unchanged (import_mod resolves via __spec__.parent and
lfx.components._importing remains a core helper).

Dep parity: uv.lock diff has zero package additions/removals (all specs
already resolved via langflow-base[complete]).

Also: 56 append-only migration entries (14 classes x 4, zero ambiguities);
component_index.json regenerated 300->286 components / 60->50 modules (exactly
the moved set); detect-secrets baseline re-keyed; 45 bundles now discover at
@official with 69 components; shims verified (azure/xai/litellm/deepseek);
ruff clean.

* fix(ci): nightly bundle rename follows [extras] refs and self-refs

Two gaps in update_bundle_versions.py around extras suffixes, both fatal
or silently wrong for the first nightly carrying the lfx-bundles
metapackage:

- update_root_pyproject_for_bundle's dep regex required the version
  specifier immediately after the name, so "lfx-bundles[all]>=1.0,<2.0"
  (a MAIN dep) was left unrewritten while the workspace member was
  renamed to lfx-bundles-nightly; uv lock then tries to resolve stable
  lfx-bundles from PyPI, where it does not exist. The same gap silently
  left "lfx-docling[local]>=0.1.0" optional-dep refs pointing at the
  stable distribution. The regex now tolerates an [extras] group and
  carries it into the replacement.

- rename_bundle_pyproject skipped self-referencing extras, so the
  metapackage's generated `all` extra kept 45 "lfx-bundles[<provider>]"
  members after the rename, pulling the stable distribution (same
  lfx_bundles import package, install collision) once published. Self
  refs now follow the rename, idempotently.

Tests drive the real script module, mirroring test_bundle_lfx_pin.py.
This closes the PR-2 audit item flagged when the metapackage was
introduced. The canonical-pre-release cutover would retire the nightly
rename entirely; until it lands, the rename must be correct.

* feat(bundles): graduate partner set to standalone lfx-<provider> packages (#13573)

* feat(bundles): graduate partner set to standalone lfx-<provider> packages

Extracts the five partner/flagship providers into manifest-shipping
distributions: lfx-openai, lfx-anthropic, lfx-amazon, lfx-datastax,
lfx-cohere. Each ships extension.json (lfx.compat ["1"]), a
langflow.extensions entry-point, an lfx>=1.10.0,<2.0.0 pin, and is wired into
the root workspace marker blocks. Zero flow impact by the bundle-name
invariant: ext:<provider>:<Class>@official ids are unchanged.

Mechanics:
- adopts the five-phase scripts/migrate/port_bundle.py from
  feat/bundle-mass-extraction (handles shared-base moves, consumer rewrites,
  surgical index updates, migration-table appends) and runs it per partner
  with --migration-release 1.11.0
- datastax's shared lfx.base.datastax moves into lfx_datastax.base; its
  backend unit tests move to src/bundles/datastax/tests (74 pass); its ruff
  per-file-ignore and its check_component_env_writes ALLOWLIST entry travel;
  10 repo consumers rewritten (incl. the vector_store_rag starter project)
- amazon_bedrock_converse.py legacy langflow.* imports rewritten to the lfx.*
  equivalents (thin re-export aliases; behavior-identical) pre-extraction
- runtime deps pinned from langflow-base's extras / direct deps (wrapper-
  guaranteed SDKs stay transitive, parity-exact); --remove-base-extra dropped
  the openai/anthropic/cohere/aws extras from langflow-base[complete] in favor
  of the bundle pins; the astradb extra stays (also used outside datastax)
- in-tree dirs replaced with marker shims pointing at lfx_<p>.components.<p>
  (skipped by the in-tree walk; legacy from lfx.components.<p> imports keep
  working); lfx/components/__init__.py entries kept, consistent with the
  consolidated providers
- validate.py: accept classes whose base is a *derived* Component base
  (LCVectorStoreComponent / LCToolComponent / ...) -- they inherit the
  class-level outputs declaration and only override the output method, which
  the AST-only check could not see; bare Component subclasses keep the strict
  build/outputs requirement (+ regression test)
- BUNDLE_API.md changelog entries added for the branch's surface changes: the
  lfx.bundles manifest-less discovery group + precedence tier +
  bundle-discovery-malformed code (from the foundation PR) and the validator
  acceptance above
- 100 append-only migration entries (20 classes x 5 partners x 4 shapes);
  component_index.json surgically updated 300->280 components / 60->55
  modules, sha256 recomputed and verified
- detect-secrets baseline re-keyed for moved partner files

Dep parity: lock diff adds only the five lfx-* package names; no third-party
package added or removed. One benign transitive re-resolution: anthropic SDK
0.105.2 -> 0.109.0 (allowed by langchain-anthropic's range on both sides).

Verified: all five register at @official via the installed-manifest tier
(extension_id/distribution = lfx-<p>, 20 components); manifest-less
lfx_bundles unchanged (35 bundles, disjoint); all 5 shims import; engine-safe;
`lfx extension validate` passes for all five; 450 extension unit tests, 60
pilot-upgrade migration tests, 74 moved datastax tests pass; ruff clean.

* fix(bundles): shim lfx.base.datastax so stored-flow imports keep resolving

The datastax graduation moved lfx.base.datastax into lfx_datastax.base,
but saved flows and starter templates (Hybrid Search RAG,
TemplateAssistant) embed `from lfx.base.datastax.astradb_base import
AstraDBBaseComponent` inside their stored component code fields, which
is re-executed verbatim at flow build time. Without a shim that import
raises ModuleNotFoundError and the flows fail to build
(test-starter-projects red).

Mirror the lfx.components shim contract: module-aliasing to
lfx_datastax.base, narrow except that only translates a missing bundle
(transitive dep failures re-raise untouched), marker-tagged single-file
dir, removed at M4 together with the components shims.

* chore(bundles): bounded version ranges for curated lfx-* packages (#13576)

langflow's pyproject is the release-coordination point now that lfx and the
lfx-* bundles release independently. Every curated lfx-* dependency declares a
BOUNDED range (>=A,<B), never an exact pin -- exact pins in reusable library
metadata create resolver conflicts for downstreams that depend on a different
version. Exact pins for reproducibility live in uv.lock and the release-build
manifests.

- the 4 pilots + 5 graduated partners: >=0.1.0 -> >=0.1.0,<1.0.0 (bundles
  follow their own 0.1.x cadence)
- lfx-bundles[all]>=1.0,<2.0 unchanged (versions with the metapackage
  contract)
- the three lfx-docling[...] optional-dependency refs bounded the same way
- policy documented inline in the bundle-deps marker block; the cross-bundle
  CI matrix verifies the ranges resolve together across the supported lfx axis

Verified: uv lock resolves with zero package/version changes (bounds are
metadata-only today); the nightly rename's dep regex already matches the
bounded form.

* test(bundles): lock the in-tree shim contract + breakage audit (#13577)

* chore(bundles): bounded version ranges for curated lfx-* packages

langflow's pyproject is the release-coordination point now that lfx and the
lfx-* bundles release independently. Every curated lfx-* dependency declares a
BOUNDED range (>=A,<B), never an exact pin -- exact pins in reusable library
metadata create resolver conflicts for downstreams that depend on a different
version. Exact pins for reproducibility live in uv.lock and the release-build
manifests.

- the 4 pilots + 5 graduated partners: >=0.1.0 -> >=0.1.0,<1.0.0 (bundles
  follow their own 0.1.x cadence)
- lfx-bundles[all]>=1.0,<2.0 unchanged (versions with the metapackage
  contract)
- the three lfx-docling[...] optional-dependency refs bounded the same way
- policy documented inline in the bundle-deps marker block; the cross-bundle
  CI matrix verifies the ranges resolve together across the supported lfx axis

Verified: uv lock resolves with zero package/version changes (bounds are
metadata-only today); the nightly rename's dep regex already matches the
bounded form.

* test(bundles): lock the in-tree shim contract + breakage audit

Locks the five-point contract for the 50 lfx-bundles import shims under
lfx/components/ (45 metapackage + 5 graduated partners):

  1. first line is the `# lfx-bundles-shim` marker (keys the in-tree walk's
     double-registration skip);
  2. a shim dir is exactly one __init__.py -- no impls, no deps;
  3. sys.modules module-aliasing (submodule trees resolve);
  4. bundle missing -> actionable ModuleNotFoundError whose wording is part
     of the contract ("pip install lfx-bundles" / "pip install lfx-<p>");
  5. a missing *transitive* dep re-raises untouched.

Test shape (env-adaptive by design -- the lfx test env prunes the venv, so
nothing here assumes bundles are installed):
- source-level sweep of every real shim, parameterized (no imports);
- hermetic mechanism tests against a synthetic shim + synthetic target
  (alias-resolves / locked-message / transitive-reraise);
- the walk-skip detector and the marker sweep must agree exactly;
- one adaptive live test on lfx.components.tavily exercising whichever
  branch the running env is in.

106 tests pass in the pruned lfx env (where the live test exercises the
bare-engine locked-message branch).

Breakage audit (gh code search, recorded in the PR): ~20-25 public repos
reference lfx.components.*; the majority are langflow forks (vendored tree,
unaffected by packaging). Genuine library consumers exist and the moved
providers have real surface (openai: 11 external repos, datastax: 17) --
covered by the shims wherever bundles are co-installed (every
`pip install langflow`). Decision: deprecation warnings NOT warranted now;
revisit at M4 (shim removal), where one loud minor release before removal
is the right shape.

* test(bundles): extend the shim contract sweep to lfx.base shims

The datastax graduation moved lfx.base.datastax into lfx_datastax.base
and left a module-aliasing shim behind (stored flow code fields embed
the legacy import and are re-executed verbatim at build time). Sweep
lfx/base for marker shims with the same source-level contract as the
components sweep: one-file stub, module-aliasing to the bundle's base
subpackage, narrow name-checked except, locked install message.

* docs(bundles): install shapes, override rule, bundle_api_version (#13578)

Documents the deliberately small user-facing rule surface of the metapackage
split (1.11):

- extensions-overview.mdx: the two install stories (pip install langflow =
  everything, same as today; pip install lfx = engine only, bring your own
  bundles); the bundle-name-is-identity invariant; the current package table
  (lfx-bundles metapackage + the 5 partner packages + the 4 pilots); the
  legacy-import shim behavior with the locked ModuleNotFoundError wording and
  the migration recipe for direct lfx users (switch to lfx[bundles] or pin
  lfx-<provider> packages); the override rule ("ship a manifest -- a manifest
  always wins", with the bundle-shadowed warning); bundle_api_version as the
  single compat number (lfx.compat ["1"]; manifest-less packages ride their
  PEP 508 lfx pin) and the no-version-arithmetic rule.
- deployment-lfx-compatibility.mdx: lfx[bundles] as the headless/serverless
  deployment footnote (engine + lfx-bundles[all]; slimmer per-provider
  alternative; intentionally no lfx[all]).

Both files verified MDX-safe (no unbackticked angle brackets).

* fix(bundles): stack-review fixes — symlink containment, idempotency, docs accuracy (#13579)

* chore(bundles): bounded version ranges for curated lfx-* packages

langflow's pyproject is the release-coordination point now that lfx and the
lfx-* bundles release independently. Every curated lfx-* dependency declares a
BOUNDED range (>=A,<B), never an exact pin -- exact pins in reusable library
metadata create resolver conflicts for downstreams that depend on a different
version. Exact pins for reproducibility live in uv.lock and the release-build
manifests.

- the 4 pilots + 5 graduated partners: >=0.1.0 -> >=0.1.0,<1.0.0 (bundles
  follow their own 0.1.x cadence)
- lfx-bundles[all]>=1.0,<2.0 unchanged (versions with the metapackage
  contract)
- the three lfx-docling[...] optional-dependency refs bounded the same way
- policy documented inline in the bundle-deps marker block; the cross-bundle
  CI matrix verifies the ranges resolve together across the supported lfx axis

Verified: uv lock resolves with zero package/version changes (bounds are
metadata-only today); the nightly rename's dep regex already matches the
bounded form.

* test(bundles): lock the in-tree shim contract + breakage audit

Locks the five-point contract for the 50 lfx-bundles import shims under
lfx/components/ (45 metapackage + 5 graduated partners):

  1. first line is the `# lfx-bundles-shim` marker (keys the in-tree walk's
     double-registration skip);
  2. a shim dir is exactly one __init__.py -- no impls, no deps;
  3. sys.modules module-aliasing (submodule trees resolve);
  4. bundle missing -> actionable ModuleNotFoundError whose wording is part
     of the contract ("pip install lfx-bundles" / "pip install lfx-<p>");
  5. a missing *transitive* dep re-raises untouched.

Test shape (env-adaptive by design -- the lfx test env prunes the venv, so
nothing here assumes bundles are installed):
- source-level sweep of every real shim, parameterized (no imports);
- hermetic mechanism tests against a synthetic shim + synthetic target
  (alias-resolves / locked-message / transitive-reraise);
- the walk-skip detector and the marker sweep must agree exactly;
- one adaptive live test on lfx.components.tavily exercising whichever
  branch the running env is in.

106 tests pass in the pruned lfx env (where the live test exercises the
bare-engine locked-message branch).

Breakage audit (gh code search, recorded in the PR): ~20-25 public repos
reference lfx.components.*; the majority are langflow forks (vendored tree,
unaffected by packaging). Genuine library consumers exist and the moved
providers have real surface (openai: 11 external repos, datastax: 17) --
covered by the shims wherever bundles are co-installed (every
`pip install langflow`). Decision: deprecation warnings NOT warranted now;
revisit at M4 (shim removal), where one loud minor release before removal
is the right shape.

* docs(bundles): install shapes, override rule, bundle_api_version

Documents the deliberately small user-facing rule surface of the metapackage
split (1.11):

- extensions-overview.mdx: the two install stories (pip install langflow =
  everything, same as today; pip install lfx = engine only, bring your own
  bundles); the bundle-name-is-identity invariant; the current package table
  (lfx-bundles metapackage + the 5 partner packages + the 4 pilots); the
  legacy-import shim behavior with the locked ModuleNotFoundError wording and
  the migration recipe for direct lfx users (switch to lfx[bundles] or pin
  lfx-<provider> packages); the override rule ("ship a manifest -- a manifest
  always wins", with the bundle-shadowed warning); bundle_api_version as the
  single compat number (lfx.compat ["1"]; manifest-less packages ride their
  PEP 508 lfx pin) and the no-version-arithmetic rule.
- deployment-lfx-compatibility.mdx: lfx[bundles] as the headless/serverless
  deployment footnote (engine + lfx-bundles[all]; slimmer per-provider
  alternative; intentionally no lfx[all]).

Both files verified MDX-safe (no unbackticked angle brackets).

* fix(bundles): review fixes — symlink containment, idempotency, docs accuracy

Fixes from the multi-agent stack review (kept as a separate PR so the
reviewed PRs' diffs stay frozen for QA's independent pass):

- _bundles_root.py: directory-level symlink containment — a provider dir
  that resolves outside the bundle root is skipped with a typed
  bundle-discovery-malformed warning, mirroring the seed-directory walk's
  is_within rule (+ regression test). Anchors the trust boundary to the
  installed package tree.
- consolidate_bundles.py: migration-append idempotency guard — entries are
  deduped on full content so a partially-failed earlier run cannot duplicate
  rows on re-run (table stays append-only).
- langflow-base pyproject: documented WHY the aws extra is deliberately
  retained after the lfx-amazon graduation (boto3 also backs the server's
  own S3 storage backend and lfx's S3 ingestion — review suggested removing
  it; that would regress S3 storage support).
- extensions-overview.mdx: note that `lfx extension list` shows
  manifest-shipping extensions only; manifest-less packages load at startup
  but are not listed.

93 loader+migration tests pass (incl. the new symlink test); ruff clean.

* refactor(bundles): stabilize import_mod as a public BUNDLE_API utility

The lazy-import helper that bundle packages call from their
__getattr__-based __init__.py files lived at lfx.components._importing -- an
internal path with no stability contract, imported by 35 separately-installed
bundle __init__ files (30 lfx-bundles providers + the 5 graduated partners).

- canonical home is now lfx.utils.lazy_import.import_mod (code unchanged);
  lfx.components._importing re-exports it so in-tree callers and any external
  code on the old path keep working
- all 35 bundle-side imports rewritten to the stable path
- BUNDLE_API.md: surface table entry + changelog (additive)
- contract tests: re-export identity, both call forms, the AttributeError
  conversion

Verified: identity holds across both paths; lazy __init__ loads work through
the new path for both bundle families; 110 tests pass; ruff clean.

* fix(bundles): tombstone the broken legacy ZepChatMemory build method (#13580)

build_message_history targeted the zep-python v1 SDK (ZepClient +
zep_python.langchain.ZepChatMessageHistory); both were removed in
zep-python 2.x and the zep extra pins 2.0.2, so the method has been
unable to run for as long as the pin has existed -- its ImportError
guard misleadingly told users to 'pip install zep-python' (already
installed). The component is legacy=True with helpers.Memory as its
designated replacement, so rather than hand-write a new integration
against the 2.x SDK, the method now raises a clear RuntimeError
pointing at the Message History component.

Flow identity is preserved: class/component name, display_name,
description, inputs and the memory output are byte-identical, so saved
flows keep loading, i18n locale keys are unchanged, and
migration_table.json needs no edits. New bundle tests pin the stub
contract (identity, actionable error, no zep_python import).

* test(bundles): extend the shim contract sweep to lfx.base shims

The datastax graduation moved lfx.base.datastax into lfx_datastax.base
and left a module-aliasing shim behind (stored flow code fields embed
the legacy import and are re-executed verbatim at build time). Sweep
lfx/base for marker shims with the same source-level contract as the
components sweep: one-file stub, module-aliasing to the bundle's base
subpackage, narrow name-checked except, locked install message.

* test(bundles): extras-drift guard for the lfx-bundles metapackage

Risk-2 of the metapackage split: the generated `all` extra and the
per-provider extras must never drift by hand-edit, or `pip install
langflow` silently loses a provider's deps. Guard the four invariants:
extras <-> provider dirs (PEP 685-normalized), `all` == the exact
self-ref set, normalized keys collision-free, and the metapackage
provider set disjoint from the graduated partner distributions.

* fix(extension): symlink-escaped providers reject with path-escape, matching the documented contract

The lfx.bundles provider containment check (stack-review symlink fix)
emitted bundle-discovery-malformed, whose template and hint describe a
broken entry-point declaration.  The changelog's path-safety entry
already documents symlink escapes as path-escape on every other
discovery path; use the same code here.  Also resolves the semantic
merge conflict with the per-mode code split (the removed
_malformed_error location kwarg).

* fix(tests): repoint lfx tests off moved providers; complete provider fallback map

CI fail-fast had been masking these: the LFX test job runs in an
engine-only env where openai/anthropic/chroma are now bundle-package
shims, so every test that used them as the example category failed on
all Python versions (3.14 just reported first), and the backend Group 2
leg failed collecting test_lfx_bundles_extras.py on Python 3.10.

- flow_requirements: complete _PROVIDER_PACKAGE_FALLBACKS for the nine
  moved model providers (OpenAI, Anthropic, Amazon Bedrock, Groq,
  Google Generative AI, SambaNova, IBM watsonx.ai, Ollama + existing
  Azure OpenAI). In engine-only installs MODEL_PROVIDERS_DICT registers
  only in-tree providers, so the dynamic source-inspection path cannot
  resolve moved providers; the static fallbacks keep lfx run/serve
  requirements inference working. A dict hit still wins.
- test_dynamic_imports / test_import_utils: use composio (still-in-tree
  lazy category with its SDK absent in the bare env) as the example
  category instead of openai/anthropic/chroma; patch import_module at
  its canonical home lfx.utils.lazy_import.
- flow-builder tests (build_flow_from_spec, flow_builder_tools,
  propose_field_edit): specs use LanguageModelComponent (in-tree)
  instead of OpenAIModel.
- test_lfx_bundles_extras: tomli fallback for Python 3.10 (tomllib is
  stdlib 3.11+; pytest guarantees tomli on <3.11).

Full lfx unit suite: 4550 passed. Backend extras+pin tests: 24 passed.

* fix(tests): repoint MCP client-server tests off graduated OpenAIModel key

Same fail-fast-masked class as 14dfa7f268, backend side: since the
partner graduation, extension components register in /api/v1/all under
their namespaced ids (ext:openai:OpenAIModelComponent@official), so the
bare 'OpenAIModel' key these tests hardcoded no longer exists on any
Python version (the 'py3.14-only' Group 2 failure was the only leg that
ran; 3.10 was cancelled by fail-fast). The MCP feature itself is fine —
search returns the namespaced ids and add_component accepts them.

- Use LanguageModelComponent (in-tree, stable bare key; SecretStr
  api_key, real_time_refresh fields, advanced 'stream', LanguageModel
  output) for the redaction/configure/search/describe/spec tests.
- configure_dynamic_field: model_name -> api_key (the refresh field on
  LanguageModelComponent).
- prompt-template-variables spec: drop the model node — server-side
  validation builds the graph and a model without an API key fails its
  build; the test's subject is the dynamic {var} fields.

Full file: 69 passed.

* Update test_mongodb_atlas.py

* Frontend tests

* Frontend test updates

* fix: restore loading.py and tableAutoCellRender to base-branch state

Commit 9df9475615 ('Frontend test updates') accidentally reverted both
files to their pre-#9902 state, removing the no_env_fallback contract in
load_from_env_vars / update_table_params_with_load_from_db_fields, the
defensive params.pop('code') in build_custom_component, and the
localValue/updateGlobalVariableCell handling in tableAutoCellRender.
The base branch's regression tests (test_loading_no_env_fallback.py,
test_loading_custom_component_code_param.py, tableAutoCellRender
index.test.tsx) were kept, so the reversion failed them.

* ci(bundles): make freeze gate reachable on PRs; review follow-ups

- Move the freeze-components job from lint-py.yml (workflow_call/
  workflow_dispatch only -- never invoked, so the gate could not block a
  PR) into extension-migration-checks.yml, which already triggers on
  pull_request for src/lfx/src/lfx/components/**. Add the gate's script
  and baseline to the path filter.
- Add timeout-minutes to the cross-bundle-test jobs so a hung install or
  test leg cannot occupy a runner for the 360-minute default.
- _discover_shimmed_component_dirs: read only the first line of each
  __init__.py instead of the whole file -- the shim marker contract is
  line 1 (enforced by test_shim_source_contract), and this prevents a
  non-shim file with the marker after leading blank lines from being
  misclassified and silently skipped from the in-tree walk.

* fix(bundles): restore legacy-name resolution for ext components; sync starter projects

Saved flows reference moved providers by their legacy palette identity --
either the bare class name (TavilySearchComponent) or the component's
name attribute (AstraDB, Chroma, needle, OllamaModel). After the provider
move neither resolved to a current template on the backend:

- lfx.utils.component_aliases now derives the bare class name (and its
  Component-stripped form) from ext:<bundle>:<Class>@<slot> keys,
  mirroring the frontend's getTemplateAliases. Ext templates carry
  name=None / _type='Component', so the key is the only source.
- _decorate_template_with_extension now stamps template['name'] with the
  component's legacy identity (name attr, falling back to class name).
  In-tree palette entries expose it as the dict key; ext entries are
  keyed by namespaced id, so without this both alias maps lose the only
  bridge from node types like 'AstraDB' to the current template.

Without these, the starter-project updater could not match moved-provider
nodes, so their embedded code stayed one whitespace change behind the
bundle sources -- every such flow showed the update-all banner, and
because tool-mode nodes' saved outputs ([component_as_tool]) differ from
the template's natural outputs the update was flagged breaking, opening
the confirmation modal instead of toasting 'successfully updated':
deterministic timeouts in the starter-project Playwright specs (shards
29/30/36/37). Regenerated the 13 affected starter projects with
scripts/ci/update_starter_projects.py (idempotent; code_hash hex strings
in the regenerated JSONs are detect-secrets false positives).

Also converts test_groq_integration.py to the captured-module
patch.object pattern (same module-identity trap as test_mongodb_atlas:
dotted-name @patch can land on a different materialization of the moved
module than the one that defined the class, leaving the real
get_groq_models in the instance's globals -- order-dependent failure in
unit-test Group 5).

* fix(bundles): keep importable module paths in starter projects

The starter-project updater syncs node metadata from the live templates
(NODE_FORMAT_ATTRIBUTES includes 'metadata'), and ext components report
their runtime sys.modules namespace there (_lfx_ext.official.<p>.<m>) --
a path that only exists inside a running extension loader. Persisting it
broke test_template_field_order_matches_component in both template jobs
('No module named _lfx_ext'): the test imports metadata.module to
instantiate the component, and the migration table likewise keys on the
legacy lfx.components.<provider> form that the bundle shims keep
importable.

_merge_node_metadata now preserves the node's stored module when the
live template's is a runtime _lfx_ext.* path, while still syncing the
rest of the metadata (e.g. dependencies). Starter projects regenerated
from the pre-pollution state with the fixed updater: the only diff vs
the previous regen is the 20 module lines reverting to the legacy
importable form. Regression tests added for both merge directions.

* feat: add Oracle integration (#13502)

* Add Oracle Integration

* [autofix.ci] apply automated fixes

* Fix styleUtils problem

* fix(oracledb): address review feedback

* fix(oracledb): apply formatting

* Fix component index

* refactor(oracledb): convert Oracle integration to lfx-oracle extension bundle

Port the in-tree lfx.components.oracledb provider into a standalone
Extension Bundle distribution (src/bundles/oracle, dist lfx-oracle,
package lfx_oracle, bundle name 'oracle') per src/bundles/PORTING.md:

- Move the four components + connection helper into the bundle; deps
  (oracledb, langchain-oracledb, langchain-community) move from the
  langflow-base[oracledb] extra into the bundle's pyproject
- Add migration-table entries mapping legacy bare class names and
  lfx.components.oracledb.* import paths to ext:oracle:<Class>@official
- Add test_pilot_oracle_upgrade.py integration coverage (mirrors ibm)
- Move backend unit tests into src/bundles/oracle/tests with bundle
  import paths; drop the empty ComponentTestBaseWithoutClient
  version-fixture scaffolding (component is new in 1.11.0)
- Declare explicit outputs on OracleVectorStoreComponent (mirrors
  LCVectorStoreComponent) so lfx extension validate passes
- Regenerate component index (oracledb category removed) and uv.lock;
  wire workspace member/dep/source in root pyproject
- Frontend sidebar entry renamed oracledb -> oracle; docs page renamed
  to bundles-oracle following the extracted-bundle convention

---------

Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Eric Hare <ericrhare@gmail.com>

* feat(lfx): fail-fast dependency preflight for `lfx run` (#13680)

Detect a flow's required third-party packages before the graph load and
fail fast with an actionable `pip install ...` message when any are
missing, instead of a deep ModuleNotFoundError mid-build. Addresses the
'export a flow, run it on a bare pip install lfx, it fails' confusion
introduced by the bundle split.

- flow_requirements: add find_missing_dependencies / _is_installed /
  format_missing_dependencies_error (best-effort, reuses the existing
  flow analyzer).
- run.base: _preflight_dependencies helper + check_dependencies gate in
  run_flow (JSON flows only; .py uses PEP 723; fail-open on analysis
  errors so a runnable flow is never blocked by the check itself).
- CLI: --check-dependencies/--no-check-dependencies on `lfx run`.
- tests: unit (find_missing_dependencies/format) + integration (run_flow
  + _preflight_dependencies).

* feat(bundles): move spider + toolguard provider components into lfx-bundles

Two provider components were still living inside otherwise-core directories
(file-level stragglers from the metapackage split). Both are portable
(lfx-only imports) and gated by dedicated optional SDKs, so they belong in
the lfx-bundles metapackage rather than lfx core:

- SpiderTool (langchain_utilities/spider.py) -> lfx_bundles/spider/, dep
  spider-client (`spider` extra).
- PoliciesComponent + policies/ submodule (models_and_agents) ->
  lfx_bundles/toolguard/, dep toolguard (`toolguard` extra).

Changes:
- Move the files into lfx_bundles/<name>/; rewrite policies' absolute
  self-imports (lfx.components.models_and_agents.policies -> lfx_bundles.toolguard.policies).
- Drop SpiderTool / PoliciesComponent from the core langchain_utilities /
  models_and_agents lazy-import registries.
- Add `spider` and `toolguard` extras to lfx-bundles pyproject + the generated
  `all` aggregate; uv.lock re-resolved (uv lock --check clean).
- Append migration entries (4 per class) mapping the historical
  lfx.components.* paths to ext🕷️SpiderTool@official /
  ext:toolguard:PoliciesComponent@official so saved flows migrate.
- Regenerate component_index.json (both classes drop out of core; 263
  components / 45 categories).
- Repoint the toolguard policies tests to lfx_bundles.toolguard (60 pass).

The toolguard deadlock-warmer (_warm_circular_imports) is unaffected: it warms
toolguard.runtime directly and runs single-threaded before the unified
component fan-out that loads bundle components.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* chore: auto-bake note keys and regenerate backend locales/en.json [skip ci]

* feat(bundles): consolidate cleanlab, twelvelabs, jigsawstack into lfx-bundles

First directory-level batch of the core-tail consolidation (validates the
consolidate_bundles.py path end-to-end after the spider/toolguard file-level
extractions). All three are flat, lfx-only, single-SDK providers:

- cleanlab    -> lfx_bundles/cleanlab    (cleanlab-tlm)   3 components
- twelvelabs  -> lfx_bundles/twelvelabs  (twelvelabs)     7 components
- jigsawstack -> lfx_bundles/jigsawstack (jigsawstack)   11 components

Via scripts/migrate/consolidate_bundles.py --apply: moves each dir into the
metapackage + leaves a fail-soft shim, merges per-provider extras and regen
`all`, appends the 4-entry migration block per class (84 entries, 0 ambiguous).
Component index regenerated (45->42 categories, 263->242 core components);
uv.lock re-resolved; twelvelabs test repointed to lfx_bundles.twelvelabs (passes).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* feat(bundles): consolidate vector/datastore + community-wrapper providers into lfx-bundles

Second directory-level batch (8 providers, 12 components) via
consolidate_bundles.py --apply:

- baidu (qianfan + langchain-community), redis (redis + community),
  elastic (elasticsearch + langchain-elasticsearch + opensearch-py)
- bing, cloudflare, maritalk, searchapi, vectara (langchain-community REST wrappers)

Moves each dir into lfx-bundles + fail-soft shim, merges per-provider extras
and regen `all`, appends 4-entry migration block per class (48 entries, 0
ambiguous). Component index regenerated (42->34 categories, 242->230 core
components); uv.lock re-resolved.

Test fixups:
- Repoint baidu/elastic/searchapi backend tests to lfx_bundles.<provider>.
- test_dynamic_imports (lfx isolated suite) used the now-bundled searchapi as
  its fixture; bundles aren't installed in the engine-only lfx env, so switch
  it to the core langchain_utilities.FakeEmbeddingsComponent (same
  langchain_community optional-import path). 131 shim tests + 21 dynamic-import
  tests pass.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* [autofix.ci] apply automated fixes

* feat(bundles): consolidate homeassistant, olivya, agentql REST providers into lfx-bundles

Third batch (3 providers, 4 components) via consolidate_bundles.py --apply.
These are thin REST wrappers with no dedicated vendor SDK:

- homeassistant (requests) -> lfx_bundles/homeassistant   2 components
- olivya (httpx, lfx core)  -> lfx_bundles/olivya          1 component
- agentql (httpx, lfx core) -> lfx_bundles/agentql         1 component

Moves + shims + per-provider extras + `all` regen; 16 migration entries (0
ambiguous). Component index 34->31 categories, 230->226 components; uv.lock
re-resolved. No tests reference these providers.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* feat(bundles): consolidate google, vertexai, altk, codeagents into lfx-bundles

Fourth directory-level batch (4 providers, 14 components) via
consolidate_bundles.py. google/vertexai (langchain-google-*), altk
(agent-lifecycle-toolkit), codeagents (smolagents + OpenDsStar); platform/py
markers copied verbatim from langflow-base extras. 56 migration entries (0
ambiguous); index 31->27 categories, 226->212 components. Backend tests
repointed; starter projects embedding moved components regenerated via the
validator. SKIP=detect-secrets: only flags are regenerated code_hash hex
(false positives).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(lfx): decouple dynamic-import unit tests from soon-to-be-bundled providers

test_dynamic_imports.py and test_import_utils.py used composio/nvidia/huggingface
as the "lazy category whose SDK is absent in the engine-only lfx env" fixture.
That couples the import-mechanism tests to specific providers — they have broken
each time a provider graduated to a bundle (openai/anthropic/chroma -> composio,
now composio et al.).

Switch to stable in-tree core categories:
- files_and_knowledge.KnowledgeComponent for the "missing-dep" path (its module
  imports langchain_chroma / langflow, absent in bare lfx) — works for both the
  import_mod direct calls and the lfx.components hierarchy access.
- helpers / processing for the "imports cleanly" and "module not found" paths.

No behavior change; 43 tests pass in the isolated lfx env. This unblocks moving
composio, huggingface, nvidia, cuga into lfx-bundles without breaking these tests.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* feat(bundles): consolidate composio, huggingface, nvidia, cuga into lfx-bundles

Fifth batch (4 providers, 71 components) via consolidate_bundles.py --apply,
unblocked by the prior dynamic-import test decoupling:

- composio    -> lfx_bundles/composio    (composio + composio-langchain)  63 components
- huggingface -> lfx_bundles/huggingface (langchain-huggingface[marker], huggingface-hub, community)  2 components
- nvidia      -> lfx_bundles/nvidia      (langchain-nvidia-ai-endpoints, nv-ingest-client[py3.12], gassist[win32])  5 components
- cuga        -> lfx_bundles/cuga        (cuga; platform/py markers)  1 component

Markers copied verbatim from langflow-base extras (uv lock --check clean).
284 migration entries (0 ambiguous); component index 27->23 categories,
212->141 components.

Tests: relocate the dedicated lfx-isolated tests for the now-bundled components
to the backend suite (test_nvidia_component.py, test_cuga_session_id_fallback.py)
repointed to lfx_bundles.<provider>; repoint backend composio/huggingface/cuga
imports + a cuga patch target; pragma a fake test api_key. 43 decoupled lfx
import tests pass post-move. SKIP=detect-secrets: only flags are regenerated
component_index code_hash hex (false positives).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* [autofix.ci] apply automated fixes

* feat(bundles): consolidate langwatch evaluator into lfx-bundles

langwatch -> lfx_bundles/langwatch (1 component). The LangWatchComponent is a
pure httpx REST wrapper (its only non-core touchpoint is lfx.base.langwatch.utils,
which is also httpx-based) — the langwatch SDK extra in langflow-base is for the
tracing service, not this component, so the bundle declares no extra deps.

Moves + shim + `all` regen; 4 migration entries (0 ambiguous). Component index
23->22 categories, 141->140 components. Backend langwatch test repointed
(16 passed). SKIP=detect-secrets: only flags are regenerated component_index
code_hash hex (false positives).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* feat(bundles): consolidate FAISS, Notion into lfx-bundles (lowercase-slug support)

Uppercase source dirs need lowercase bundle slugs (BUNDLE_NAME_RE is
lowercase-only). Enhance consolidate_bundles.py with bundle_slug(provider) =
provider.lower(): the bundle dir, ext id, shim target, and extra key use the
lowercased slug, while the migration import_path keeps the historical
lfx.components.<Provider> casing so saved flows still resolve. Identity for
already-lowercase providers (no change to prior entries).

- FAISS  -> lfx_bundles/faiss  (faiss-cpu[py-split] + langchain-community)  1 component
- Notion -> lfx_bundles/notion (requests + Markdown)  8 components

test_bundle_shims.py: derive the expected bundle name by lowercasing the shim
dir name (FAISS shim aliases lfx_bundles.faiss) — 159 shim tests pass.
Relocate the lfx-isolated FAISS test to backend/vectorstores (repointed to
lfx_bundles.faiss.faiss). 36 migration entries (0 ambiguous); index 22->20
categories, 140->131 components. SKIP=detect-secrets: regenerated code_hash hex.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* feat(bundles): fold vectorstores LocalDBComponent into the chroma bundle

The lone LocalDBComponent is Chroma-backed (eager `from langchain_chroma import
Chroma`), so it belongs in the existing `chroma` bundle rather than a new one —
its deps are already covered by the chroma extra (no new deps).

- Move vectorstores/local_db.py -> lfx_bundles/chroma/local_db.py; register
  LocalDBComponent in the chroma bundle's lazy registry.
- Replace lfx.components.vectorstores with a CROSS-BUNDLE shim aliasing
  lfx_bundles.chroma (dir != bundle).
- Migration: lfx.components.vectorstores[.local_db].LocalDBComponent ->
  ext:chroma:LocalDBComponent@official (4 entries, 0 ambiguous).
- test_bundle_shims.py: add a cross-bundle map {vectorstores: chroma} so the
  shim contract validates (161 shim tests pass).
- Repoint the backend LocalDB test (import + patch target) to lfx_bundles.chroma;
  test_all_modules_importable resolves it via the shim.

Component index 20->19 categories, 131->130 components. SKIP=detect-secrets:
regenerated code_hash hex (false positives).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* [autofix.ci] apply automated fixes

* feat(bundles): remove unbundleable vlmrun + notdiamond components from lfx core

Per the core-tail audit, four providers couldn't be bundled. vlmrun (imports
`from langflow`, violating the lfx engine boundary) and notdiamond (its
`notdiamond` SDK isn't in the lockfile or any extra, so it can't run) have no
engine references, no starter flows, and no migration-table entries — they're
effectively dead in every distribution, so remove them outright.

(crewai and agentics are kept: crewai is woven into 3 starter-project example
flows, and the core Agent hard-depends on agentics.helpers.model_config — both
need deliberate upstream/refactor work, not deletion.)

- Delete src/lfx/.../components/{vlmrun,notdiamond}/ and drop them from the
  lfx.components top-level registry (TYPE_CHECKING / _dynamic_imports / __all__).
- Regenerate component index (19->17 categories, 130->128 components) and backend
  locales (orphan keys removed; extract_backend_strings --check OK).
- Repoint test_dynamic_imports.test_import_error_handling off the deleted
  notdiamond to the core processing category (43 lfx import tests pass).

SKIP=detect-secrets: only flags are regenerated component_index code_hash hex.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* test(lfx): tie flow-builder search 'returns all' to registry size

The hardcoded `count > 100` assertion predated the bundle-extraction
work, which deliberately moves non-legacy components out of lfx core
(now 66 non-legacy of 128 indexed). Assert the empty-query count equals
the full non-legacy registry instead — the durable 'no query == every
component the registry exposes' contract that survives future bundle
moves.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(bundles): deterministic component alias resolution + sidebar slug casing after bundle move

Moving agentql/composio and the long tail into lfx-bundles made two
shared-label resolutions order-dependent and left two sidebar slugs
mis-cased, surfacing as red CI on the starter-projects template tests
and the bundles-sidebar e2e.

- flatten_components_with_aliases: register every component's identity
  aliases (registry key / ext class name / name) before any component's
  display_name alias, so a component's own class identity always wins
  its key regardless of registry iteration order. A single setdefault
  pass previously let the Composio 'AgentQL'/'SerpAPI' wrappers
  (display_name) shadow the standalone components (class name) depending
  on load order, so update_starter_projects wrote the wrong
  metadata.module into the AgentQL starter node.
- Correct the AgentQL node module in News Aggregator.json and Price Deal
  Finder.json back to lfx.components.agentql.agentql_api.AgentQL.
- SIDEBAR_BUNDLES: lowercase the Notion/FAISS name slugs to match the
  lowercase bundle category names so both bundles render in the sidebar
  again (display_name and icons unchanged).
- Add order-independence regression tests for the identity/display tiers.

* test(weaviate): patch the class's own module, not a fixed string path

The Weaviate component moved into lfx-bundles, so weaviate.py is now
reachable under several module identities (the lfx.components.weaviate
shim, lfx_bundles.weaviate.weaviate, and the runtime _lfx_ext.* copy).
On Python 3.10, unittest.mock resolves a dotted patch target via a
getattr walk that follows the shim's package alias to a *different*
module object than the one the imported class actually lives in, so
patching 'lfx.components.weaviate.weaviate.WeaviateVectorStore' missed
the object the component body reads -> 'WeaviateVectorStore called 0
times' when an earlier test in the group forced the alternate identity.

Resolve the module from the imported class (sys.modules[Cls.__module__])
and patch.object it directly, bypassing dotted-string resolution. Same
fix for AuthApiKey. Order/identity-independent on all Python versions.

* test(ollama,perplexity): patch the class's own module, not a fixed string path

The Ollama and Perplexity components moved into lfx-bundles, so ollama.py
and perplexity.py are now reachable under several module identities (the
lfx.components.<x>.<x> shim, the canonical lfx_bundles.<x>.<x>, and the
runtime _lfx_ext.* ext-loader copy). On Python 3.10, unittest.mock
resolves a dotted patch target via a getattr walk that follows the shim's
package alias to a *different* module object than the one the imported
class / component body actually use, so patching e.g.
"lfx.components.ollama.ollama.ChatOllama" missed the object the body reads
-> "mock called 0 times" depending on suite order.

Resolve the module from the imported class (sys.modules[Cls.__module__])
and patch.object it directly, bypassing dotted-string resolution:
- perplexity: ChatPerplexity -> patch.object(_PERPLEXITY_MODULE, ...)
- ollama: ChatOllama / logger -> patch.object(_OLLAMA_MODULE, ...)
- ollama: ChatOllamaComponent._parse_json_response ->
  patch.object(ChatOllamaComponent, ...) (the imported class itself)
- ollama: httpx.AsyncClient.get/post -> patched on the global httpx
  (a single class object, identity-independent), dropping the shim prefix

Same fix as the Weaviate test (064069b6bb). Order/identity-independent on
all Python versions; the langchain_ollama / lfx.utils.util / socket
patches target stable modules and are left unchanged.

* fix(locales): restore en.json keys for still-shipping spider/policies components

The spider (SpiderTool) and toolguard (PoliciesComponent) components live in
lfx-bundles and still ship in the palette, but their 44 en.json keys were
pruned while all 6 translated locales (de/es/fr/ja/pt/zh-Hans) retained them.
The extractor walks only lfx.components, so bundle-only components with no
compat shim get dropped from en.json on regeneration. Left as-is, the next GP
upload/download cycle would strip ~264 existing translations (44 keys x 6
langs) for two components that are still in the palette.

Restore the 25 spidertool + 19 policies keys to re-align en.json with the
translated locales. (Durable fix — making the extractor walk src/bundles and/or
adding lfx.components shims for these two providers — is a follow-up, since
spider.py eager-imports its SDK and a shim alone won't make it walkable.)

* chore(deps): drop dead astradb/graph-retriever extras and langchain-aws from langflow-base

After datastax/amazon graduated to lfx-datastax/lfx-amazon, langflow-base still
declared astradb (langchain-astradb), graph-retriever (langchain-graph-retriever,
graph-retriever) and langchain-aws in its aws extra -- all pulled into the
complete aggregate -- though zero core .py files import them (the consuming
components moved to the bundles, which re-declare the same deps). This duplicated
the SDKs in every pip install langflow and was inconsistent with the
openai/anthropic/cohere extras already removed in this PR.

Remove the astradb and graph-retriever extras and their complete refs, and drop
langchain-aws from the aws extra (boto3 stays -- it backs langflow-base's own S3
storage, not just the amazon components). Regenerate uv.lock; closure unchanged,
the SDKs still resolve via lfx-datastax/lfx-amazon.

* ci(extension-migration): enforce component os.environ-write lint in CI

check_component_env_writes.py guards against components writing to process-global
os.environ (which bleeds one caller's credential into other requests on a warm
worker). It was wired only as a pre-commit hook, bypassable via git commit -n,
web-UI commits, or uninstalled hooks. Add it as a stdlib-only job in
extension-migration-checks.yml next to freeze-components, so the credential-bleed
guard is enforced in CI regardless of local pre-commit.

* test(bundles): cover migration-table completeness, preflight provider path, frozen gate

Three test-coverage gaps from the metapackage split, all green on the current tree:
- migration-table completeness: parametrized sweep asserting every lfx_bundles
  provider component class (196 across 72 providers) is reachable via
  migration_table.json as an ext:<provider>:<Class>@ target, so a provider moved
  into a bundle without a table entry fails CI instead of silently breaking saved
  flows. Includes a negative control.
- dependency preflight provider path: drives find_missing_dependencies and
  _preflight_dependencies through a node with a model/agent_llm provider field so
  the _PROVIDER_PACKAGE_FALLBACKS path (the only working provider-dep source in
  engine-only installs) is exercised.
- frozen-components gate: unit tests for check_components_frozen.py covering
  additions (fail), removals (allowed), and the init/pycache/dotdir and
  comment-parsing rules.

* chore: auto-bake note keys and regenerate backend locales/en.json [skip ci]

* fix(locales): make spider/toolguard i18n keys survive regen via shims

The earlier restore of the 44 spider/policies en.json keys was reverted by the
gp-backend-check bot: it auto-regenerates en.json on every push to this PR (the
PR-level path filter matches the bundle moves under lfx/components) and the
bundle-blind extractor could not reproduce the keys.

Make them durable the way every other moved provider does:
- Add lfx.components.spider and lfx.components.toolguard '# lfx-bundles-shim'
  packages re-pointing to lfx_bundles.{spider,toolguard}, so the extractor (which
  walks lfx.components) emits SpiderTool/PoliciesComponent keys again.
- Lazy-import spider's SDK (from spider.spider import Spider moved into crawl())
  so SpiderTool is importable for extraction/registration without spider-client,
  matching the lazy pattern weaviate/zep already use (toolguard was already lazy).
- Add spider/toolguard to scripts/ci/frozen_component_dirs.txt so the
  additions-only freeze gate accepts the new shim dirs.
- Restore the 44 keys to en.json.

Verified the extractor now reproduces exactly the 44 keys (25 spidertool + 19
policies), so the bot keeps them instead of dropping them. Bundle components are
dynamic-only (not in component_index.json), so no index regen is needed.

* chore(bundles): declare astrapy in lfx-datastax; align lfx-ibm manifest version

- lfx-datastax imports astrapy directly (astrapy.DataAPIClient/Database in the
  astra components) but only pulled it transitively via langchain-astradb.
  Declare it explicitly, pinned to match langflow-base's astrapy extra
  (>=2.1.0,<3.0.0).
- lfx-ibm's extension.json reported version 0.1.0 while pyproject.toml is 0.1.1;
  the loader uses the manifest version, so bump it to 0.1.1 to match.
Regenerate uv.lock (closure unchanged; astrapy was already resolved
transitively).

* refactor(migrate): guard consolidate against subdir data loss; fix alias docstring

- consolidate_bundles.move_provider copies only top-level *.py but rmtree's the
  whole source tree, so a provider with a subpackage would have its subdir
  silently destroyed and never migration-mapped. Refuse with a clear error
  instead of half-moving (port_bundle.py handles nested layouts). No current
  PROVIDER_DEPS entry has a subdir, so this is a defensive guard.
- component_aliases: the order-independence docstrings overclaimed. Scope the
  guarantee to identity-beats-display (and real-key-beats-alias) and note that
  two components contributing the same identity alias remain first-wins by
  iteration order.

* ci(gp-backend): regen en.json on src/bundles/** edits

The gp-backend-check bot regenerates locales/en.json on PR changes, but its
path filter only watched src/lfx/src/lfx/components/**. Bundle component
strings reach core en.json through the lfx/components/<provider> import shims
(e.g. spider, toolguard), so editing a bundle's display_name/description/info
under src/bundles/** would not re-trigger the extractor and en.json could
silently drift from the translated locales. Add src/bundles/** to the trigger
so bundle-side string edits keep en.json aligned; regen is diff-gated, so
non-string bundle edits remain a no-op.

* fix(bundles): Policies in Models & Agents core; Spider in bundles sidebar

Policies (toolguard) component:
- Move PoliciesComponent + policies/ helpers out of the lfx-bundles
  consolidated metapackage back into lfx/components/models_and_agents core
  (restores the main/release-1.11.0 layout); register it in the category
  __init__ and restore the core-path tests.
- toolguard stays an optional langflow-base extra (langflow-base[toolguard],
  pulled in via [complete]); the component imports it lazily, so a bare lfx
  install still loads/inspects it.
- Remove the lfx-bundles toolguard provider (dir + extra + [all] entry), the
  lfx core import shim, and the frozen_component_dirs.txt entry; drop the 4
  ext:toolguard migration entries; regenerate component_index (128 -> 129)
  and uv.lock.

Spider:
- Add spider to SIDEBAR_BUNDLES so it renders in the Bundles section like
  arXiv instead of falling into the default categories (it was missing from
  the list while peer lfx-bundles providers were present).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* docs(bundles): fix error-tier mirror claim, list-time discovery note, metapackage README

Addresses three documentation nits from the PR #13563 re-review:

- errors.py: the lfx.bundles diagnostics comment claimed a full mirror of the
  inline tier's split, but only the duplicate case matches. name-invalid and
  root-unreadable are warning-only here where their inline counterparts are
  hard errors (ok=False). Tighten the comment to match actual behavior.
- BUNDLE_API.md: note that manifest-less lfx.bundles providers are invisible to
  `lfx extension list` (discover_all_extensions walks only installed + seed;
  load_lfx_bundles_extensions runs in the startup component-loading path), so
  the precedence story isn't read as a list-time property.
- lfx-bundles/README.md: the bulk move has landed on this branch (71 providers,
  populated per-provider extras, generated `all`), so drop the "empty skeleton /
  available later" framing and document the final state.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(bundles): surface pip-install guidance for missing SDKs on the server run path

A flow loaded through the UI/API on an engine-only `lfx` (no bundles)
failed deep in the vertex build with a bare
`Error creating class. ModuleNotFoundError(No module named 'langchain_chroma')`.
The actionable install guidance only existed on the `lfx run` preflight
and the legacy import shim; the server run path
(simple_run_flow -> run_graph_internal -> graph.arun()) had neither.

Both vertex-build error formatters (lfx's `format_exception_message`, used
by graph.arun(), and langflow's, used by the build/chat endpoints) now
detect a plain ModuleNotFoundError in the cause chain and attach the same
`pip install <package>` guidance the CLI gives, mapping the missing import
to its PyPI distribution via the existing requirements resolver. Bundle-shim
"components moved to ..." messages are left untouched (their message does
not start with "No module named"), so graduated providers stay actionable.

* fix(bundles): surface curated shim message over its raw cause in exception formatter

get_causing_exception walked to the deepest __cause__, so a graduated
provider's curated "components moved to ..." ModuleNotFoundError (raised
`from` the raw "No module named 'lfx_<x>'") was unwrapped past it. The
formatter then regenerated bare "pip install lfx-<x>" guidance, dropping
the curated text — including the "or pip install langflow, which bundles
it" clause.

Both get_causing_exception copies (lfx + langflow) now stop at the first
ModuleNotFoundError whose message is not a raw "No module named ..." (the
curated shim error), and module_not_found_hint surfaces that curated
message verbatim. Plain SDK import errors are unchanged.

Adds regression tests reproducing the real 3-level cause chain on both
formatters; the old test used a truncated chain that hid this.

* fix(bundles): cover OpenRouter/IBM WatsonX in Orchestrate requirements detection (#13732)

fix(bundles): cover OpenRouter/IBM WatsonX in wxO requirements detection

The bundle split removed the langchain provider SDKs from lfx's own
dependency tree, which makes flow→requirements detection load-bearing for
Watsonx Orchestrate deployments: the runner installs only `lfx` + the
packages `generate_requirements_from_flow` emits, so a provider missing
from the resolution tables yields an empty requirements set and the
deployed flow fails to import its model class.

Two unified-selectable providers were unresolved:
- OpenRouter (runs on ChatOpenAI) was in no resolution table -> [].
- "IBM WatsonX" (catalog casing) only matched the legacy "IBM watsonx.ai"
  key; langchain-ibm landed only by an embedding-path coincidence.

Add both to _PROVIDER_PACKAGE_FALLBACKS (keyed by the exact catalog
strings) so every provider in MODEL_PROVIDER_METADATA resolves.

Tests:
- TestBundleSeparationOrchestrate in test_flow_requirements.py: catalog
  coverage guard, OpenRouter/IBM regressions, and split-contract guards
  (bundle dists and langchain imports must stay out of lfx's tree).
- End-to-end artifact tests in test_watsonx_orchestrate.py asserting a
  bundle-provider flow emits the langchain SDK (not lfx-openai/lfx-bundles)
  and that requirements.txt pins lfx as its first line.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

* Backend format

* Format again

* [autofix.ci] apply automated fixes

* fix(bundles): pin datastax ruff line-length to 120 to stop format tug-of-war

The datastax bundle's pyproject.toml declares a [tool.ruff.lint.per-file-ignores]
table, which makes ruff treat the directory as a standalone config root. It never
restated line-length, so ruff fell back to its default of 88 there. As a result
`make format_backend` (per-directory config discovery) wrapped lines to 88 while
the pre-commit ruff-format hook (run with --config pyproject.toml, forcing the
repo-wide 120) unwrapped them back, leaving the two formatters fighting forever.

Restate line-length = 120 so both config-resolution paths agree. Use the explicit
line-length rather than extend=../../../pyproject.toml to avoid pulling in the
root lint ruleset, which the bundle intentionally does not enforce.

* fix(bundles): relocate notdiamond + vlmrun into lfx-bundles

Both providers were dropped during the bulk long-tail move with no bundle
and no compatibility shim, removing them from the registry,
component_index, migration table and locales. That breaks saved flows
using the NotDiamond Router or VLM Run Transcription nodes (vlmrun ships
in langflow[complete]), contradicting the zero-user-facing-change goal.

Relocate both the same way as the other long-tail providers:
- move implementations into src/bundles/lfx-bundles/src/lfx_bundles/
- restore lfx.components.{notdiamond,vlmrun} marker shims
- re-register both in lfx/components/__init__.py
- add per-provider extras (vlmrun -> vlmrun[all]) and the 'all' aggregate
- add migration_table targets ext:{notdiamond,vlmrun}:<Class>@official
- restore the 32 en.json strings; refresh uv.lock

* docs(lfx): correct dependency-preflight scope to lfx run only

The preflight is wired into 'lfx run'; 'lfx serve' has its own
module-not-found hint at request time and does not call it.

---------

Co-authored-by: Elif Sema Balcioglu <elifsemabalcioglu@gmail.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-authored-by: github-actions[bot] <github-actions[bot]@users.noreply.github.com>
2026-06-23 19:37:10 +00:00

25 KiB

Bundle API

Stable surface that Langflow Extension Bundles consume. Every public symbol listed below is part of the contract: changes to its name, signature, semantics, or visibility require a coordinated version bump and a ## Changelog entry.

This document is paired with the integer BUNDLE_API_VERSION declared in lfx.extension.manifest. Manifests declare the contract versions they support via lfx.compat: ["1"]; a bundle that does not list str(BUNDLE_API_VERSION) is rejected at install time with version-constraint-unsatisfied.

CI gate: any PR that modifies a file containing an in-scope surface MUST add a ## Changelog entry describing the change. The CI guard scripts/migrate/check_bundle_api_changelog.py enforces this. Pure-internal refactors that preserve every public symbol's name and signature do not require a changelog entry, but reviewers should be skeptical.


Surface (v0)

Component base class

Symbol Source
Component lfx.custom.custom_component.component.Component
Component.build() (declared on subclasses) call site of every loaded bundle module
Component.inputs declarative input list
Component.outputs declarative output list
Component.display_name / Component.description / Component.icon / Component.documentation metadata read by the palette
Component.name optional override of the registry class name

Inputs

Symbol Source
Input (base) lfx.io
MessageTextInput / MultilineInput / SecretStrInput lfx.io
IntInput / FloatInput / BoolInput lfx.io
DropdownInput / TabInput lfx.io
DictInput / NestedDictInput lfx.io
FileInput / LinkInput lfx.io
HandleInput lfx.io

Outputs

Symbol Source
Output lfx.io

Schema types

Symbol Source
Data lfx.schema.data
DataFrame lfx.schema.dataframe
Message lfx.schema.message

Lazy-import helper (consumed by bundle __init__.py files)

Symbol Source
import_mod(attr_name, module_name, package) lfx.utils.lazy_import (canonical; lfx.components._importing re-exports it for backward compatibility)

Manifest contract (consumed by the loader)

Symbol Source
Manifest schema (extension.json / [tool.langflow.extension]) lfx.extension.manifest.ExtensionManifest
BundleRef (one entry in bundles[]) lfx.extension.manifest.BundleRef
LfxCompat (declared as manifest.lfx) lfx.extension.manifest.LfxCompat
BUNDLE_API_VERSION (the integer this lfx ships) lfx.extension.manifest
EXTENSION_SCHEMA_URL / SCHEMA_VERSION lfx.extension.manifest

Slot vocabulary: official (installed pip distributions and seed directories) and extra (paths declared in LANGFLOW_COMPONENTS_PATH). Component IDs at runtime are ext:<bundle>:<Class>@<slot>.

Discovery + loading entry points

Symbol Source
load_extension(root) lfx.extension.loader
load_installed_extensions() lfx.extension.loader
discover_inline_bundles() lfx.extension.loader
discover_installed_extensions() / discover_seed_extensions() / discover_all_extensions() lfx.extension.discovery
LoadedComponent lfx.extension.loader (frozen dataclass; what the registry stores)
LoadResult lfx.extension.loader
SLOT_OFFICIAL / SLOT_EXTRA lfx.extension.loader

Reload pipeline

Symbol Source
reload_bundle(registry, bundle_name) lfx.extension.reload
BundleRegistry lfx.extension.bundle_registry
BundleRecord lfx.extension.bundle_registry
ReloadInProgressError lfx.extension.bundle_registry
POST /api/v1/extensions/{id}/bundles/{name}/reload langflow.api.v1.extensions

Errors

Symbol Source
ExtensionError lfx.extension.errors
ExtensionErrorCollection lfx.extension.errors
format_extension_error(error) lfx.extension.errors
ERROR_CODES (frozenset of every typed code) lfx.extension.errors

The full kebab-case discriminant set is the contract — adding a code is backward-compatible; removing or renaming a code is a breaking change and requires a BUNDLE_API_VERSION bump.

Validate / authoring CLI

Symbol Source
validate_extension(root, *, execute_imports=False) lfx.extension.validate
ValidateReport lfx.extension.validate
lfx extension validate (CLI) lfx.cli._extension_commands
lfx extension schema (CLI) lfx.cli._extension_commands
lfx extension init (CLI) lfx.cli._extension_commands
lfx extension dev (CLI -- registers a local path and execs langflow run) lfx.cli._extension_commands
lfx extension list (CLI) lfx.cli._extension_commands
lfx extension reload (CLI) lfx.cli._extension_commands
register_dev_extension / unregister_dev_extension (Python API) lfx.extension.dev_registry

Migration

Symbol Source
Migration-table file src/lfx/src/lfx/extension/migration/migration_table.json
MigrationEntry lfx.extension.migration.schema
MigrationTable lfx.extension.migration.schema
migrate_flow_payload(payload, table) lfx.extension.migration.rewrite
MIGRATION_SCHEMA_VERSION lfx.extension.migration.schema

Out of scope (v0)

These are reserved in the manifest schema and produce a typed field-deferred-in-this-milestone error if set; they are NOT part of the v0 contract:

  • services — bundle-declared service factories
  • routes — bundle-mounted HTTP routes
  • hooks — bundle-declared lifecycle hooks
  • starter_projects — bundle-shipped starter flows
  • userConfig — bundle-declared user-config schema
  • Multi-bundle manifests (bundles list with length > 1)

Pilot bundle: lfx-duckduckgo

The shipped LE-1023 pilot is duckduckgo, extracted into the standalone distribution lfx-duckduckgo under src/bundles/duckduckgo/ with its own pyproject.toml. langflow's own pyproject.toml declares lfx-duckduckgo>=0.1.0 as a regular dependency so a flat pip install langflow continues to ship the bundle as before.

Why this bundle:

  • Single component (DuckDuckGoSearchComponent) in a single file (duck_duck_go_search_run.py).
  • Zero git churn over the last six months.
  • Modern Component base class (no LCToolComponent legacy).
  • No authentication required — failure mode is a single failed request, not a paid-API outage.
  • Class name is globally unique across src/lfx/src/lfx/components/**, so the bare-name migration entry is allowed by check_bare_names.py.

The runtime half of the M1 proof-of-delivery gate (save a flow on pre-migration Langflow, upgrade, confirm it loads AND runs identically) lives in the dogfood checklist at src/bundles/duckduckgo/M1_DOGFOOD_CHECKLIST.md; the deserialize half is covered by src/lfx/tests/integration/extension/test_pilot_duckduckgo_upgrade.py.


Changelog

v0 (this release)

  • Initial surface enumerated above. Frozen as BUNDLE_API_VERSION = 1.
  • BundleRegistry.write_locked() exposed as a public context manager so the reload pipeline can hold the registry write lock across both the sys.modules swap and the BundleRecord install. Concurrent readers can no longer observe new modules paired with the old record. No change to the addressable component contract.
  • HTTP reload endpoint (POST /api/v1/extensions/{id}/bundles/{name}/reload) returns 422 Unprocessable Entity for structural failures (broken bundle, missing source path, name mismatch) instead of 200 OK with ok=false. Body is {...primaryError, result: ReloadResult} so the full typed result is preserved under the FastAPI detail envelope. 409 Conflict for reload-in-progress is unchanged.
  • CLI table updated to remove the obsolete dev register / dev unregister / dev list subcommands; the actual surface is extension dev <path> plus the Python helpers register_dev_extension / unregister_dev_extension.
  • MigrationTable.ambiguous_bare_names added. Each entry is {name, candidates: [list of canonical IDs]} and registers a bare class name that exists in 2+ bundles. The deserializer now surfaces component-name-ambiguous (with the candidate targets) for any bare name listed here, instead of falling through to the generic component-not-found-with-hint. Seeded with the canonical regression cases (MergeDataComponent, SplitTextComponent, SubFlowComponent). check_bare_names.py now verifies every Component class found in 2+ bundle folders has a matching marker, so a future bundle move that introduces a new ambiguity is caught at PR time.
  • Router-trust CI guard broadened to scan every .py under src/backend/base/langflow/api/** and src/lfx/src/lfx/**; a new file that mounts an APIRouter(prefix=".../extensions...") is auto-detected and checked for forbidden install/uninstall/registry-mutation handlers. Authors of files with non-literal prefixes can opt in via a # router-trust: in-scope marker.
  • Router-trust guard rewritten to use AST-based cross-file resolution. A forbidden handler in module A is now caught when module B mounts A's router via parent.include_router(child, prefix=".../extensions..."), and the same applies transitively across multi-hop include_router chains. An imported router that cannot be statically resolved is ignored (the guard never flags routes it cannot prove reachable from /extensions); routes co-located with an in-scope router ARE flagged.
  • check_migration_append_only.py now compares ambiguous_bare_names alongside entries. A marker may not be removed once published, and its candidates list may only grow -- shrinking it would silently regress flows from component-name-ambiguous to component-not-found-with-hint.
  • Router-trust guard now resolves dotted attribute references in include_router and decorators. include_router(child.api.router, prefix="/extensions") after import child.api (and the import child.api as alias; alias.router shape) are caught -- not just from child.api import router as child_router. The parser flattens any Name/Attribute chain, and the resolver walks imports of either kind (from M import N and import M, with or without an asname) back to the source file.
  • Router-trust guard's relative-import resolver is now __init__.py-aware. Inside a package, from .child import Y anchors at the package itself (level=1 -> pkg); inside a regular module pkg.foo it anchors at the parent package (level=1 -> pkg). The arithmetic differs because __init__.py's file module IS the package, while pkg/foo.py's file module is pkg.foo. The resolver tracks is_package and decrements level by one for __init__.py files so both shapes resolve correctly.
  • Code-review hardening pass across the extension subsystem. No public symbol's name or signature changed; this entry covers behavioural tightening that bundle authors and operators should be aware of:
    • Path-safety contract honored on every discovery path. DiscoveredExtension records emitted from discover_installed_extensions / discover_seed_extensions now run the same resolve-and-relative_to containment check that validate_extension performs. A symlinked bundles[0].path or a symlinked seed subdirectory that escapes the extension root is now rejected with path-escape before reaching the loader, instead of slipping through to exec_module(). The shared primitive lives at lfx.extension._paths.is_within; every walker (loader, validator, seed discovery, inline-bundle discovery) uses the same function and the same SKIP_DIR_NAMES.
    • --execute-imports env allowlist. The validator's --execute-imports subprocess now inherits an explicit allowlist (PATH, LANG, LC_*, SYSTEMROOT, TMPDIR, TZ, Python locale + encoding vars) instead of denylisting only LANGFLOW_*/LFX_*. Cloud / CI credentials (AWS_*, OPENAI_API_KEY, GITHUB_TOKEN, ...) no longer propagate into untrusted bundle import. The CLI / module docs re-frame this pass as best-effort hygiene lint, not a sandbox.
    • AST hygiene lint widened. _find_top_level_io now flags exec, eval, __import__, compile as top-level primitives and importlib.import_module / importlib.__import__ as dotted-name primitives. Still best-effort literal-name matching; trivially bypassable by obfuscation, and documented as such.
    • Reload swap is non-destructive. _swap_sys_modules now builds the staging->prod rename map before any sys.modules mutation, snapshots popped old modules into a recovery map, and restores them on any mid-swap exception. The length-mismatch tripwire on zip(strict=True) no longer leaves the prod namespace shredded. A new typed code, reload-class-retag-failed, is appended to ReloadResult.warnings when cls.__module__ cannot be retagged so the empty-palette-after-reload regression leaves a trail instead of silently failing.
    • Cross-source bundle-name collision. load_installed_extensions now detects two distributions with different canonical names but identical bundle.name (which would silently clobber each other at _lfx_ext.official.<name>.*) and emits a typed duplicate-bundle-name error on the loser, dropping its components. BundleRegistry.install_bundle additionally logs a WARNING when an existing record is replaced by a record from a different source_path (catches collisions the upstream precedence resolver missed).
    • Reload endpoint off event loop. POST /api/v1/extensions/{id}/bundles/{name}/reload now invokes reload_bundle via asyncio.to_thread so slow or large bundle imports do not freeze the worker for other in-flight requests. The wire contract (status codes, body shape) is unchanged.
    • Stable typed-error code rename. multi-bundle-deferred-in-this-milestone is renamed to the stable multi-bundle-unsupported. The old code is retained in ERROR_CODES as a deprecated alias for one milestone for log scrapers. Three new codes are added to ERROR_CODES: duplicate-bundle-name (see above), reload-class-retag-failed (see above), and reload-transport-error (CLI-side connectivity failure, previously misreported as reload-source-missing).
    • Discovery preserves "unreadable" vs "absent" distinction. _pyproject_declares_extension now propagates OSError so a permission failure on a pyproject that might declare an extension surfaces as manifest-unreadable instead of being silently dropped as "no extension here".
    • Dev registry corruption is logged. _read_state now distinguishes file absent (silent, legitimate empty registry), file present but unreadable (WARNING), and file present but corrupt JSON / wrong shape (WARNING with detail). The state file is written with mode 0600 so a hostile third-party process cannot inject an extension path into the developer's next langflow run.
    • Entry-point predicate avoids module-level side effects. _entry_point_loads_to_component now consults importlib.util.find_spec first and only falls through to ep.load() when the spec lookup is insufficient. The except BaseException was narrowed to except Exception so SystemExit / KeyboardInterrupt are no longer swallowed at filter time.
    • Frontend reload-success warnings surfaced. The reload route's ReloadResult.warnings (non-empty on success) now reach the user via a notice toast in addition to the green success toast. Wire shape unchanged; this is a UI fix that consumes existing payload fields.
    • Internal-only file split. sys.modules surgery primitives moved to lfx.extension.reload_swap; load_installed_extensions / load_seed_extensions moved to lfx.extension.loader._startup. Both are re-exported from their previous import paths so external imports are unchanged.
    • Editable installs are discovered via the entry-point fallback. _distribution_manifest_path now falls back to the langflow.extensions entry-point group when dist.files only surfaces dist-info/ entries (the pip install -e / uv pip install -e case). The entry-point value is resolved via importlib.util.find_spec -- which runs import-system finders but never executes the module body -- and the resulting package directory is scanned for extension.json or a [tool.langflow.extension] pyproject. Wheel installs are unaffected: the fallback only fires when the primary dist.files scan finds no manifest. Previously, editable-installed bundles were silently dropped by lfx extension list and the registry, even though the bundle pyproject already declared the entry-point.
    • Reload CLI: --bundle is optional; --all is implemented. lfx extension reload <ext_id> now resolves the bundle name from local discover_all_extensions when --bundle is omitted; explicit --bundle still wins for cases where the local install is not visible to the running server. lfx extension reload --all iterates every locally-discovered bundle, POSTs reload to each, and exits non-zero if any reload fails (previously hard-errored as "not yet wired"). --all is mutually exclusive with a positional id / --bundle (exit 2). The HTTP wire contract (POST /api/v1/extensions/{id}/bundles/ {name}/reload per-bundle) is unchanged; this is a CLI-only surface change.
  • User-scoped extension events. Bundle lifecycle events (bundle_reloaded, bundle_reload_failed, flow_migrated, extension_error) now publish to a per-user keyspace (user:<user_id>) instead of the shared "global" bucket so flow-migration and reload payloads cannot leak across users via the poll endpoint.
    • reload_bundle gains an optional keyword-only user_id: str | None = None argument. When supplied, bundle_reloaded / bundle_reload_failed events are emitted to keyspace user:<user_id>; None (CLI / authless dev) keeps the legacy "global" emission. Existing positional callers are unaffected.
    • POST /api/v1/extensions/{id}/bundles/{name}/reload now resolves the authenticated user and threads its id into reload_bundle, so every reload triggered via HTTP is published to that user's keyspace. Wire contract (status codes, body shape) unchanged.
    • GET /api/v1/extensions/events drops its client-supplied keyspace query parameter. The endpoint derives the keyspace from the authenticated user server-side, so an authenticated client can no longer poll another user's keyspace. Frontends that polled without keyspace (the in-tree consumer) are unaffected; third-party callers that explicitly passed keyspace=... will now receive 422 from FastAPI's strict parameter validation.
  • Reload event payload aligned with ReloadResult. Both bundle_reloaded and bundle_reload_failed events now carry the full ReloadResult.to_dict() envelope (ok, bundle, reload_id, components_added, components_removed, components_changed, warnings, errors) instead of a hand-rolled subset. Polling clients can now (a) detect body-only edits via components_changed instead of mis-reporting them as "no source changes detected", and (b) surface a failed reload's errors[0].message instead of degrading to a generic "check server logs" fallback. HTTP response shape unchanged.
  • GET /api/v1/extensions/events rejects keyspace explicitly. Previously the endpoint accepted but silently ignored any client-supplied keyspace query parameter (server-derived from the authenticated user since the prior entry). Silent drop masked client bugs that assumed the value had effect. The route now returns 422 Unprocessable Entity with a typed extension-events-keyspace-forbidden error envelope when the parameter is present. extension-events-keyspace-forbidden is added to ERROR_CODES (additive; codes-as-contract semantics preserved). In-tree polling clients that never sent the parameter are unaffected.
  • Manifest-less lfx.bundles discovery (metapackage split, 1.11). A distribution may declare [project.entry-points."lfx.bundles"] whose value is an importable package; each immediate subdirectory is loaded as one bundle at the @official slot with no extension.json (the langchain-community model; these directories are not inputs to lfx extension validate -- the validator requires an extension.json or [tool.langflow.extension] manifest and reports manifest-not-found without one). load_lfx_bundles_extensions is exported from lfx.extension (additive). Discovery precedence for cross-source bundle-name collisions becomes installed > seed > lfx_bundles > dev > inline -- manifest always wins, so a manifest-shipping lfx-<provider> shadows the same-named provider in a metapackage with the existing bundle-shadowed warning (graduation requires no lockstep release). A metapackage provider whose name is already claimed by an installed or seed source is never imported -- all @official sources share the _lfx_ext.official.<bundle>.* sys.modules namespace, so importing the losing copy would overwrite the winner's live modules; the skipped copy carries the same typed bundle-shadowed diagnostic (on errors, matching the resolver) so filtering by code never mixes severities. Namespace packages are walked across all portions, and resolved roots are deduplicated by path so duplicate declarations never self-shadow. This precedence is a startup-time property: load_lfx_bundles_extensions runs in the component-loading path, not in discover_all_extensions, so lfx extension list -- which walks only the installed and seed (manifest-bearing) sources -- does not enumerate manifest-less metapackage providers. A live lfx.bundles provider is therefore invisible at list-time, and its shadowing against a manifest-shipping package is resolved only when the server assembles the _lfx_ext.official.<bundle>.* namespace at startup. New warning-only codes added to ERROR_CODES (additive), one per discovery failure mode and none of which abort startup: bundle-discovery-malformed (declaration does not resolve to an importable package directory -- including a parent package whose __init__ raises during find_spec), bundles-provider-name-invalid (provider folder is not a valid bundle name), bundles-root-unreadable (root cannot be enumerated), and duplicate-lfx-bundles-provider (same provider name in more than one root; first wins). Manifest-less bundles bypass the version-constraint-unsatisfied API-version gate by construction (no manifest to carry lfx.compat); install-time compatibility rides on the metapackage's PEP 508 lfx>=X,<Y pin instead. Manifest-less records register with manifestless=True (additive field on LoadResult / BundleRecord) and are not hot-reloadable: the reload pipeline refuses them with the new typed code reload-manifestless-unsupported (additive in ERROR_CODES) instead of failing manifest-not-found; pick up metapackage changes by upgrading the distribution and restarting the process.
  • import_mod promoted to a stable public home. The lazy-import helper that bundle packages call from their __getattr__-based __init__.py files moved from the internal lfx.components._importing to lfx.utils.lazy_import and is now part of the BUNDLE_API surface (name, signature, and semantics contract-stable). lfx.components._importing re-exports it unchanged, so existing in-tree and third-party callers are unaffected (additive).
  • Validator accepts inherited entry-points. validate_extension no longer emits build-method-missing for a Component subclass whose base is a derived Component base (name ends with Component, e.g. LCVectorStoreComponent / LCToolComponent): such classes inherit the class-level outputs declaration and only override the output method, which the AST-only check cannot resolve across modules. Direct subclasses of the root Component keep the strict inline build-or-outputs requirement. Surfaced by the partner graduations (lfx-datastax et al.); previously-passing bundles are unaffected (strictly fewer false positives).