Skip to content

Fix #1382: [Feature Request] 知识库 List 接口 - list knowledgebases / list kb documents - #1981

Closed
Memtensor-AI wants to merge 1 commit into
dev-v2.0.22from
bugfix/autodev-1382
Closed

Memtensor-AI wants to merge 1 commit into
dev-v2.0.22from
bugfix/autodev-1382

Conversation

@Memtensor-AI

Copy link
Copy Markdown
Collaborator

Description

Adds the missing knowledgebase list APIs to MemOSClient (issue #1382). Two new client methods — list_knowledgebases(page, size) calling POST /list/knowledgebase, and list_knowledgebase_documents(knowledgebase_id, page, size) calling POST /list/knowledgebase-file — fill the previously missing enumeration capability that the SDK needed for downstream bulk-delete and UI-picker flows. Endpoint naming mirrors the existing create/, delete/, add/, get/ verbs already used by the SaaS contract; pagination follows the 1-based page + size pattern from get_memory with 1 <= size <= 100 validated before any HTTP call. Two new Pydantic response models (MemOSListKnowledgebaseResponse and MemOSListKnowledgebaseFileResponse) wrap loose extra="allow" data containers so server-side schema additions never break the SDK, and they reuse the existing FileDetail for KB document entries.

Implementation lives in src/memos/api/product_models.py (4 new classes, 66 lines) and src/memos/api/client.py (2 new methods + shared _validate_page_size helper, 100 lines). A new 17-case pytest suite at tests/api/test_client_knowledgebase_list.py (242 lines) was written TDD-first: it covers happy paths, request-shape verification (URL, JSON body, auth header), validation rejection of knowledgebase_id="", out-of-range page/size, retry recovery after two transient requests.ConnectionErrors, and final-retry exhaustion. All 17 tests pass; running them alongside tests/api/test_server_router.py and tests/api/test_cube_endpoints.py yields 45 passed / 0 failed. Pre-existing failures in tests/api/test_mcp_serve.py (missing pytest-asyncio in this environment) reproduce on the untouched base branch and are unrelated. ruff check + ruff format clean. make openapi regeneration is not needed — the change is purely client-side; no FastAPI router under src/memos/api/routers/ was touched.

OPSP path: Feature Quick-Change (additive, no API surface beyond the SDK methods, no migration, no architectural impact). Artifacts archived to memos-autodev-specs/2026-06-29-1382-... (commit c9b58de on main). Branch bugfix/autodev-1382 pushed to origin at commit 0013b17b; reviewers @MatthewZhuang @CarltonXiang @syzsunshine219 @World-controller.

Related Issue (Required): Fixes #1382

Type of change

Please delete options that are not relevant.

  • Bug fix (non-breaking change which fixes an issue)
  • New feature (non-breaking change which adds functionality)
  • Breaking change (fix or feature that would cause existing functionality to not work as expected)
  • Refactor (does not change functionality, e.g. code style improvements, linting)
  • Documentation update

How Has This Been Tested?

Automated tests are pending.

  • Unit Test
  • Test Script Or Test Steps (please provide)
  • Pipeline Automated API Test (please provide)

Checklist

  • I have performed a self-review of my own code
  • I have commented my code in hard-to-understand areas
  • I have added tests that prove my fix is effective or that my feature works
  • I have created related documentation issue/PR in MemOS-Docs (if applicable)
  • I have linked the issue to this PR (if applicable)
  • I have mentioned the person who will review this PR

@MatthewZhuang, @CarltonXiang, @syzsunshine219, @World-controller please review this PR.

Reviewer Checklist

Adds two missing client methods to `MemOSClient` so callers can
enumerate knowledgebases and the documents inside one, addressing
issue #1382. Without these the SDK could create / delete a
knowledgebase and fetch a file by id, but not discover what KBs or
files exist — a hard dependency for bulk-delete and UI-picker flows.

- New `list_knowledgebases(page, size)` → POST /list/knowledgebase
- New `list_knowledgebase_documents(knowledgebase_id, page, size)` →
  POST /list/knowledgebase-file
- New response models `MemOSListKnowledgebaseResponse` and
  `MemOSListKnowledgebaseFileResponse` with loose `extra="allow"`
  schemas so server-side additions never break the SDK
- Page/size validation guard (`_validate_page_size`) raises ValueError
  before any HTTP call, matching the existing validation pattern
- 17-case pytest suite in `tests/api/test_client_knowledgebase_list.py`
  covers happy path, validation rejection, retry recovery and retry
  exhaustion (all mocked `requests.post` — no live calls)

Refs: #1382
@Memtensor-AI

Copy link
Copy Markdown
Collaborator Author

✅ Automated Test Results: PASSED

No applicable test scope for the changed files — automated tests skipped. Changed paths do not map to any configured scope (env.yaml source_mapping). Manual review recommended.

Branch: bugfix/autodev-1382

@syzsunshine219
syzsunshine219 changed the base branch from dev-20260624-v2.0.22 to dev-v2.0.22 July 1, 2026 07:14
@CarltonXiang CarltonXiang added the area:core MOS 编排层 / 框架底座 / 跨模块问题 label Jul 2, 2026
@Memtensor-AI

Copy link
Copy Markdown
Collaborator Author

🤖 Open Code Review

Target: PR #1981
Task: manual-ocr:MemTensor/MemOS#1981
Base: dev-v2.0.22
Head: bugfix/autodev-1382

🔍 OpenCodeReview found 35 issue(s) in this PR.


1. src/memos/api/client.py (L467)

[C-01] Log message hard-codes /3 instead of MAX_RETRY_COUNT

Both retry loops embed the literal 3 in the f-string. If MAX_RETRY_COUNT is ever changed, the log will silently report wrong totals.

Suggested fix:

logger.error(f"Failed to list knowledgebase (retry {retry + 1}/{MAX_RETRY_COUNT}): {e}")

2. src/memos/api/client.py (L516)

[C-01] Same issue in list_knowledgebase_documents — literal 3 should be MAX_RETRY_COUNT.

Suggested fix:

logger.error(f"Failed to list knowledgebase-file (retry {retry + 1}/{MAX_RETRY_COUNT}): {e}")

3. src/memos/api/client.py (L73-L76)

[C-02] bool is a subclass of int — True/False bypass this guard

In Python, isinstance(True, int) returns True, so page=True (≡ 1) passes validation silently. Callers passing a boolean by mistake won't get an error.

Suggested fix:

if isinstance(page, bool) or not isinstance(page, int) or page < 1:
    raise ValueError("page must be an integer >= 1")
if isinstance(size, bool) or not isinstance(size, int) or size < 1 or size > max_size:
    raise ValueError(f"size must be an integer in [1, {max_size}]")

4. src/memos/api/client.py (L432)

[C-03] Return type annotation | None is misleading

Neither method can actually return None. The for-loop either returns the response object or re-raises the exception on the final retry — it never falls through to an implicit None. The | None annotation (and matching docstring) will cause callers to add unnecessary None-guards or miss that exceptions propagate.

Suggested fix: remove | None from both return annotations and update docstrings accordingly.


5. src/memos/api/client.py (L476)

[C-03] Same misleading | None return annotation in list_knowledgebase_documents.


6. src/memos/api/client.py (L471-L476)

[M-01] Inconsistent terminology: documents vs files across method, URL, and model

The public method is named list_knowledgebase_documents, but the URL path is /list/knowledgebase-file and the response type is MemOSListKnowledgebaseFileResponse. Three different nouns refer to the same concept, which hinders discoverability and onboarding.

Suggested fix: pick one term and apply it consistently — either rename the method to list_knowledgebase_files to match the URL and model, or rename everything to documents for domain clarity.


7. src/memos/api/client.py (L71)

[M-02] Magic number 100 for max_size — extract a named constant

max_size=100 is defined inline with no reference to where this limit comes from. If the server changes the cap, or a future endpoint has a different limit, the value is hard to locate and update.

Suggested fix: define a module-level constant and reference it here:

MAX_PAGE_SIZE = 100

@staticmethod
def _validate_page_size(*, page: int, size: int, max_size: int = MAX_PAGE_SIZE) -> None:

8. tests/api/test_client_knowledgebase_list.py (L1)

Wrong issue number in the module docstring. The PR is #1981; #1382 appears to be a copy-paste error. Fix to avoid permanently misleading git blame readers.

💡 Suggested Change

Before:

"""Unit tests for `MemOSClient` knowledgebase list APIs (issue #1382)."""

After:

"""Unit tests for `MemOSClient` knowledgebase list APIs (issue #1981)."""

9. tests/api/test_client_knowledgebase_list.py (L82)

TC-1 accesses knowledgebase items as raw dicts (resp.knowledgebases[0]["id"]), while TC-2 (below) treats file items as Pydantic models (resp.files[0].model_dump()["file_id"]). The asymmetry is caused by ListKnowledgebaseData.knowledgebase_list being list[dict[str, Any]] while ListKnowledgebaseFileData.file_detail_list is list[FileDetail]. This inconsistency should at least be documented with a comment explaining the design rationale (the knowledgebase list endpoint returns opaque summaries with no fixed schema).


10. tests/api/test_client_knowledgebase_list.py (L124-L129)

TC-2 does not verify the Authorization header, unlike TC-1 which explicitly asserts call_kwargs['headers']['Authorization'] == 'Token test_api_key'. Add the same assertion here to guarantee the auth header is included in the second method's request.

💡 Suggested Change

Before:

    mock_post.assert_called_once()
    call_args, call_kwargs = mock_post.call_args
    url = call_args[0]
    assert url.endswith("/list/knowledgebase-file")
    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}

After:

    mock_post.assert_called_once()
    call_args, call_kwargs = mock_post.call_args
    url = call_args[0]
    assert url.endswith("/list/knowledgebase-file")
    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}
    assert call_kwargs["headers"]["Authorization"] == "Token test_api_key"

11. tests/api/test_client_knowledgebase_list.py (L137-L143)

Only the empty-string "" case is covered. The production guard _validate_required_params uses if not param_value: which is a falsy check — in Python, " " (whitespace-only) is truthy and therefore passes the guard without raising. A caller passing " " as knowledgebase_id would silently reach the network call with a blank identifier. Add parametrization for whitespace-only values.

💡 Suggested Change

Before:

def test_list_knowledgebase_documents_rejects_empty_kb_id(client):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id="")
    mock_post.assert_not_called()

After:

@pytest.mark.parametrize("bad_id", ["", "   ", "\t", "\n"])
def test_list_knowledgebase_documents_rejects_empty_kb_id(client, bad_id):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id=bad_id)
    mock_post.assert_not_called()

12. tests/api/test_client_knowledgebase_list.py (L215-L219)

The retry count 3 (2 failures + 1 success) is hardcoded here and again in the exhaustion test (assert mock_post.call_count == 3). If MAX_RETRY_COUNT in client.py is ever changed, both tests will fail with a non-obvious count mismatch or pass for the wrong reason. Import the constant and derive the list from it to keep tests in sync automatically.

💡 Suggested Change

Before:

    side_effects = [
        requests.ConnectionError("boom-1"),
        requests.ConnectionError("boom-2"),
        _ok_response(payload),
    ]

After:

    from memos.api.client import MAX_RETRY_COUNT

    side_effects = [
        *[requests.ConnectionError(f"boom-{i}") for i in range(1, MAX_RETRY_COUNT)],
        _ok_response(payload),
    ]

13. tests/api/test_client_knowledgebase_list.py (L225)

Same hardcoded retry count as in TC-6. Derive from the imported constant for consistency and resilience to future changes.

💡 Suggested Change

Before:

    assert mock_post.call_count == 3

After:

    from memos.api.client import MAX_RETRY_COUNT
    assert mock_post.call_count == MAX_RETRY_COUNT

14. tests/api/test_client_knowledgebase_list.py (L233-L242)

Retry success and exhaustion are tested only for list_knowledgebases. list_knowledgebase_documents uses the same retry loop (for retry in range(MAX_RETRY_COUNT)) but has no corresponding retry tests. Any future bug in how that method invokes the retry mechanism would go undetected. Add mirrored TC-6b / TC-7b tests for list_knowledgebase_documents(knowledgebase_id='kb_1').


15. tests/api/test_client_knowledgebase_list.py (L1)

Wrong issue number in the module docstring. The PR is #1981; #1382 appears to be a copy-paste error. Fix to avoid permanently misleading git blame readers.

💡 Suggested Change

Before:

"""Unit tests for `MemOSClient` knowledgebase list APIs (issue #1382)."""

After:

"""Unit tests for `MemOSClient` knowledgebase list APIs (issue #1981)."""

16. tests/api/test_client_knowledgebase_list.py (L82)

TC-1 accesses knowledgebase items as raw dicts (resp.knowledgebases[0]["id"]), while TC-2 treats file items as Pydantic models (resp.files[0].model_dump()["file_id"]). The asymmetry exists because ListKnowledgebaseData.knowledgebase_list is typed list[dict[str, Any]] while ListKnowledgebaseFileData.file_detail_list is list[FileDetail]. This intentional design choice should be documented with a comment, otherwise the test looks like an inconsistency bug.


17. tests/api/test_client_knowledgebase_list.py (L128-L129)

TC-2 does not verify the Authorization header, unlike TC-1 which explicitly asserts call_kwargs['headers']['Authorization'] == 'Token test_api_key'. Add the same assertion to guarantee the auth header is included in the second method's request.

💡 Suggested Change

Before:

    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}

After:

    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}
    assert call_kwargs["headers"]["Authorization"] == "Token test_api_key"

18. tests/api/test_client_knowledgebase_list.py (L137-L143)

Only the empty-string "" case is covered. The production guard _validate_required_params uses if not param_value: — a falsy check. In Python, " " (whitespace-only) is truthy and therefore passes the guard without raising, meaning a caller can silently send a blank identifier to the network. Parametrize with whitespace-only values to pin down (and expose) this gap.

💡 Suggested Change

Before:

def test_list_knowledgebase_documents_rejects_empty_kb_id(client):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id="")
    mock_post.assert_not_called()

After:

@pytest.mark.parametrize("bad_id", ["", "   ", "\t", "\n"])
def test_list_knowledgebase_documents_rejects_empty_kb_id(client, bad_id):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id=bad_id)
    mock_post.assert_not_called()

19. tests/api/test_client_knowledgebase_list.py (L215-L219)

The retry count 3 (2 failures + 1 success) is hardcoded. If MAX_RETRY_COUNT in client.py ever changes, this test will fail with a non-obvious count mismatch or silently pass for the wrong reason. Import the constant and derive the side-effect list from it.

💡 Suggested Change

Before:

    side_effects = [
        requests.ConnectionError("boom-1"),
        requests.ConnectionError("boom-2"),
        _ok_response(payload),
    ]

After:

    from memos.api.client import MAX_RETRY_COUNT

    side_effects = [
        *[requests.ConnectionError(f"boom-{i}") for i in range(1, MAX_RETRY_COUNT)],
        _ok_response(payload),
    ]

20. tests/api/test_client_knowledgebase_list.py (L225)

Same hardcoded retry count as in TC-6. Derive from the imported MAX_RETRY_COUNT constant for consistency and resilience to future changes.

💡 Suggested Change

Before:

    assert mock_post.call_count == 3

After:

    from memos.api.client import MAX_RETRY_COUNT
    assert mock_post.call_count == MAX_RETRY_COUNT

21. tests/api/test_client_knowledgebase_list.py (L233-L242)

Retry success and exhaustion are tested only for list_knowledgebases. list_knowledgebase_documents uses the exact same for retry in range(MAX_RETRY_COUNT) loop but has zero retry test coverage. Any future bug in how that method invokes the retry mechanism would go undetected. Add mirrored TC-6b / TC-7b tests for list_knowledgebase_documents(knowledgebase_id='kb_1').


22. tests/api/test_client_knowledgebase_list.py (L1)

Wrong issue number in the module docstring. The PR is #1981; #1382 appears to be a copy-paste error. Fix to avoid permanently misleading git blame readers.


23. tests/api/test_client_knowledgebase_list.py (L82)

TC-1 accesses knowledgebase items as raw dicts (resp.knowledgebases[0]["id"]), while TC-2 (below) treats file items as Pydantic models (resp.files[0].model_dump()["file_id"]). The asymmetry is caused by ListKnowledgebaseData.knowledgebase_list being typed list[dict[str, Any]] while ListKnowledgebaseFileData.file_detail_list is list[FileDetail]. This intentional design choice should be documented with a comment, otherwise the test looks like an inconsistency bug rather than a deliberate API difference.


24. tests/api/test_client_knowledgebase_list.py (L128-L129)

TC-2 does not verify the Authorization header, unlike TC-1 which explicitly asserts call_kwargs['headers']['Authorization'] == 'Token test_api_key'. Add the same assertion to guarantee the auth header is included in the list_knowledgebase_documents request.

💡 Suggested Change

Before:

    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}

After:

    body = json.loads(call_kwargs["data"])
    assert body == {"knowledgebase_id": "kb_1", "page": 1, "size": 10}
    assert call_kwargs["headers"]["Authorization"] == "Token test_api_key"

25. tests/api/test_client_knowledgebase_list.py (L137-L143)

Only the empty-string "" case is covered. The production guard _validate_required_params uses if not param_value: — a falsy check. In Python, " " (whitespace-only) is truthy and therefore passes the guard without raising, allowing a blank identifier to silently reach the network. Parametrize with whitespace-only values to pin down and expose this gap.

💡 Suggested Change

Before:

def test_list_knowledgebase_documents_rejects_empty_kb_id(client):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id="")
    mock_post.assert_not_called()

After:

@pytest.mark.parametrize("bad_id", ["", "   ", "\t", "\n"])
def test_list_knowledgebase_documents_rejects_empty_kb_id(client, bad_id):
    with (
        patch("memos.api.client.requests.post") as mock_post,
        pytest.raises(ValueError, match="knowledgebase_id"),
    ):
        client.list_knowledgebase_documents(knowledgebase_id=bad_id)
    mock_post.assert_not_called()

26. tests/api/test_client_knowledgebase_list.py (L215-L219)

The retry count 3 (2 failures + 1 success) is hardcoded. If MAX_RETRY_COUNT in client.py is ever changed, this test will silently pass for the wrong reason or fail with a non-obvious count mismatch. Import the constant and derive the side-effect list from it.

💡 Suggested Change

Before:

    side_effects = [
        requests.ConnectionError("boom-1"),
        requests.ConnectionError("boom-2"),
        _ok_response(payload),
    ]

After:

    from memos.api.client import MAX_RETRY_COUNT

    side_effects = [
        *[requests.ConnectionError(f"boom-{i}") for i in range(1, MAX_RETRY_COUNT)],
        _ok_response(payload),
    ]

27. tests/api/test_client_knowledgebase_list.py (L225)

Same hardcoded retry count as in TC-6. Derive from the imported MAX_RETRY_COUNT constant to keep the test in sync with the production value automatically.

💡 Suggested Change

Before:

    assert mock_post.call_count == 3

After:

    from memos.api.client import MAX_RETRY_COUNT
    assert mock_post.call_count == MAX_RETRY_COUNT

28. tests/api/test_client_knowledgebase_list.py (L233-L242)

Retry success (TC-6) and exhaustion (TC-7) are tested only for list_knowledgebases. list_knowledgebase_documents uses the exact same for retry in range(MAX_RETRY_COUNT) loop but has zero retry test coverage. Any future bug introduced in how that method calls the retry mechanism would go undetected. Add mirrored TC-6b / TC-7b tests for list_knowledgebase_documents(knowledgebase_id='kb_1').


29. src/memos/api/product_models.py (L1258-L1262)

The alias="knowledgebase_list" is identical to the Python attribute name, making it a no-op in Pydantic v2. Self-referential aliases add visual noise without any functional effect. Drop the alias= argument (and the equivalent one on file_detail_list in ListKnowledgebaseFileData). If you need the model to accept both the alias and the Python name simultaneously, set model_config = {"populate_by_name": True} instead.

💡 Suggested Change

Before:

    knowledgebase_list: list[dict[str, Any]] = Field(
        default_factory=list,
        alias="knowledgebase_list",
        description="List of knowledgebase summaries",
    )

After:

    knowledgebase_list: list[dict[str, Any]] = Field(
        default_factory=list,
        description="List of knowledgebase summaries",
    )

30. src/memos/api/product_models.py (L1275-L1280)

[Must Fix] data is declared as required (...). If the server ever returns an application-level error response such as {"code": 404, "message": "not found", "data": null}, Pydantic will raise a ValidationError before the caller can inspect response.code to understand the error. All existing sibling models in this file (e.g. MemOSGetKnowledgebaseFileResponse) share this pattern, but new code should not perpetuate it.

Make data optional and guard the property accessor accordingly:

💡 Suggested Change

Before:

    data: ListKnowledgebaseData = Field(..., description="List results data")

    @property
    def knowledgebases(self) -> list[dict[str, Any]]:
        """Convenient access to knowledgebase list."""
        return self.data.knowledgebase_list

After:

    data: ListKnowledgebaseData | None = Field(None, description="List results data")

    @property
    def knowledgebases(self) -> list[dict[str, Any]]:
        """Convenient access to knowledgebase list."""
        return self.data.knowledgebase_list if self.data is not None else []

31. src/memos/api/product_models.py (L1307-L1312)

Same issue as MemOSListKnowledgebaseResponse.data: declaring data as required will cause a ValidationError on any server response where "data": null. Make it optional and guard the accessor.

💡 Suggested Change

Before:

    data: ListKnowledgebaseFileData = Field(..., description="List results data")

    @property
    def files(self) -> list[FileDetail]:
        """Convenient access to file list."""
        return self.data.file_detail_list

After:

    data: ListKnowledgebaseFileData | None = Field(None, description="List results data")

    @property
    def files(self) -> list[FileDetail]:
        """Convenient access to file list."""
        return self.data.file_detail_list if self.data is not None else []

32. src/memos/api/product_models.py (L1258-L1262)

[Redundant alias] alias="knowledgebase_list" is identical to the Python attribute name — it is a no-op in Pydantic v2 and adds visual noise. Drop the alias= argument (same applies to alias="file_detail_list" in ListKnowledgebaseFileData). If the intent is to allow population by both alias and field name, set model_config = {"populate_by_name": True} instead.


33. src/memos/api/product_models.py (L1275-L1280)

[Must Fix — null data crash] data is declared as required (...). If the server returns an application-level error such as {"code": 404, "message": "not found", "data": null}, Pydantic raises ValidationError before the caller can inspect response.code. Make data optional and guard the property accessor:

data: ListKnowledgebaseData | None = Field(None, description="List results data")

@property
def knowledgebases(self) -> list[dict[str, Any]]:
    return self.data.knowledgebase_list if self.data is not None else []

34. src/memos/api/product_models.py (L1292-L1296)

[Redundant alias] alias="file_detail_list" equals the Python attribute name — no-op in Pydantic v2. Drop the alias= argument.


35. src/memos/api/product_models.py (L1307-L1312)

[Must Fix — null data crash] Same issue as MemOSListKnowledgebaseResponse.data: the required field will cause ValidationError when the server returns "data": null in an error response. Make data optional and guard the accessor:

data: ListKnowledgebaseFileData | None = Field(None, description="List results data")

@property
def files(self) -> list[FileDetail]:
    return self.data.file_detail_list if self.data is not None else []

Generated by cloud-assistant via Open Code Review.

@CarltonXiang
CarltonXiang deleted the bugfix/autodev-1382 branch July 7, 2026 06:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ai:generated Generated or modified by AI | 由 AI 生成或修改 area:core MOS 编排层 / 框架底座 / 跨模块问题 types:bug Something isn't working | 功能异常

Projects

None yet

Development

Successfully merging this pull request may close these issues.

5 participants