Skill v1.0.1

name: lint-python description: "Apply strict Python engineering rules before code is accepted." metadata: version: "1.0.1"

Procedure

1. Identify the .py files touched or under review.
2. Apply the Python policy in rules/10-python.mdc (agent-surface source; it may be rendered to the current host) together with the repository's configured linters and formatters. Treat Python as statically typed with runtime validation at boundaries: no Any, bare # noqa without specific codes, dict in place of data models, except Exception: pass, or ignored errors. If a constraint cannot be met, stop and name the blocker.
3. Run the toolchain gates:

``bash ruff format --check . ruff check . pyright . pytest -v ``

1. Report in the format below.

Output

1. Understanding
Task: [the task in your own words]
Constraints: [from context]
Ambiguities: [anything unclear; ask if critical]
 
2. Design
DATA MODELS:
[Pydantic models for boundaries]
[Dataclasses for internal]
ERROR STRATEGY:
[Exception hierarchy]
[Handling approach]
DEPENDENCIES:
[Each with justification]
 
3. Implementation
[Complete, working code: all files, all functions, all tests]
 
4. Verification
Toolchain commands: [list]
Ruff status: [PASS or specific issues and how resolved]
Pyright status: [PASS or specific issues and how resolved]
Test coverage: [percentage and what is tested]