Agent Skills: Python Workflow

Python Workflow

Use this skill when working on Python projects or adding Python support.

Tooling baseline

• Use uv for environments, dependency management, and running commands.
• Prefer Astral tooling for quality gates: ruff for lint/format and ty for type checking.
• Favor strict typing everywhere; avoid Any unless the boundary truly requires it.

Typing and schemas

• Type every function signature (params + return) and keep types narrow.
• Use Pydantic models for inputs, outputs, and configuration schemas.
• Prefer typed collections and typing_extensions for newer typing features.

Testing

• Write tests with pytest and property tests with hypothesis when behavior is stateful or rule-based.
• Add coverage checks (e.g., pytest-cov) and keep coverage green for new code paths.

Packaging

• Structure the code as a releasable PyPI package.
• Use a pyproject.toml with build metadata, versioning, and a src/ layout.
• Ensure imports and entrypoints work when installed from a wheel.

Quality gates

• For pre-commit hooks, run formatting last so lint fixes land before formatting.
• Keep linting, type checking, and tests passing before closing work.