FaceTransformMCP CLAUDE.md

FaceTransformMCP

MCP (Model Context Protocol) server for the AceDataCloud Face Transform API — seven face tools covering keypoint detection, beautification, age/gender transform, face swap, cartoonization, and liveness detection.

Project Structure

core/
  config.py     — Settings dataclass (API token, base URL, timeouts)
  server.py     — FastMCP server singleton + optional OAuth provider
  client.py     — FaceClient: async httpx wrapper for /face/* endpoints
  types.py      — Literal types (FaceAnalyzeMode, FaceModelVersion, ...)
  exceptions.py — Error classes (FaceAuthError, FaceAPIError, FaceTimeoutError)
  oauth.py      — Generic AceDataCloud OAuth 2.1 provider (PKCE)
  utils.py      — Shared response helpers (kept from template; unused by face)
tools/
  face_tools.py — All seven face tools + face_get_usage_guide
prompts/
  __init__.py   — face_guide, face_workflow_examples
tests/
  test_smoke.py — Smoke tests (settings, server import, tool registration)

Adding / updating a face endpoint

1. Source of truth — the upstream OpenAPI spec lives at https://docs.acedata.cloud/openapi/face.json (or the matching file in the Docs/ submodule).
2. Update client — add a method to FaceClient in core/client.py.
3. Add tool — register a new @mcp.tool() function in tools/face_tools.py.
4. Update prompts — extend face_guide and face_workflow_examples.
5. Update server card — append the tool description to the server_card payload in main.py.
6. Update README + CHANGELOG.
7. Add tests — at minimum a smoke test that the tool is registered.
8. Run ruff check . and pytest --cov=core --cov=tools before opening a PR.

Development

pip install -e ".[dev,test]"
pytest --cov=core --cov=tools
ruff check .

Local run

cp .env.example .env
# fill in ACEDATACLOUD_API_TOKEN
python main.py                          # stdio
python main.py --transport http         # HTTP on :8000