The Agent Skills Directory

[External Utility Usage]: The skill uses npx to download and run the Mermaid CLI tool (@mermaid-js/mermaid-cli) for validating diagram syntax. This is a standard and transparent method for ensuring documentation quality in developer environments.
[Evaluation and Testing]: Includes a comprehensive Python-based evaluation framework (run_evals.py) to verify trigger accuracy and content quality. The suite uses local fixtures and standard libraries, ensuring a safe and deterministic testing process.
[Output Grounding and Verification]: To ensure accuracy, the skill enforces a "cite-or-omit" rule where every claim must be backed by a verbatim quote from trusted documentation. This reduces the risk of generating misleading information from repository diffs or logs.
[Structured Ingestion Surface]: While the skill ingests untrusted repository data (like commit messages), it mitigates potential injection risks by utilizing a rigid template and a mandatory self-check rubric to validate the final output before it is saved.

pr-description-skill