Imported from yellowcandle/mv-face-recognition (AGENTS.md). Install upstream with npx skills add yellowcandle/mv-face-recognition. Copyright stays with the author.

MV Face Recognition - AI Agent Guidelines

Build/Test/Lint Commands

Python (Root & Backend):

uv run pytest - Run all tests (use uv as per .clinerules)
uv run pytest tests/test_specific.py::test_function - Run single test
uv run black . - Format code with Black
uv run isort . - Sort imports
uv run flake8 . - Lint code
uv run mypy . - Type checking

Frontend (Svelte):

cd frontend && npm run lint - ESLint + Prettier check
cd frontend && npm run format - Auto-format with Prettier
cd frontend && npm run check - Type check with svelte-check
cd frontend && npm run build - Production build

MVP Processor:

cd mvp-processor && npm test - Run Vitest tests
cd mvp-processor && npm run test:e2e - Playwright E2E tests
cd mvp-processor && npm run lint - ESLint check

Code Style Guidelines

Python: Follow PEP 8, use type hints, document functions with docstrings. Import order: stdlib, third-party, local (from src.core.detector import FaceDetector). Use numpy arrays for face processing, logging for errors.

TypeScript/Svelte: Use strict TypeScript, camelCase naming, reactive statements with $:, stores for state (dashboardStats, contestantsStore). Import from $lib/ for components and utilities.

Error Handling: Python - use logging module, TypeScript - proper try/catch with meaningful messages.

Use uv and pyproject.toml for Python dependency management (per .clinerules)

Install mv-face-recognition and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

mv-face-recognition

About