The Agent Skills Directory

[SAFE]: The skill implements a structured code review process with clear safety boundaries. It includes a validation script (scripts/validate_code_review.py) that uses the standard jsonschema library to ensure output quality. All external tool calls (git, gh) are standard for the intended development workflow. While the skill processes untrusted code (PRs/diffs), this is its primary function, and the structured workflow mitigates risks of indirect instruction obedience. No malicious patterns or unauthorized data access were identified.

code-review