The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill follows best practices by using standard development tools (git, gh) and ensuring that any code fixes are explicitly reviewed and approved by the user through a confirmation step.
[PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface as it ingests untrusted code content from external pull requests and local diffs. (1) Ingestion points: Pull request data fetched via gh pr diff and local changes via git diff. (2) Boundary markers: The skill uses descriptive prompts to separate instructions from the code content but does not employ hardened delimiters. (3) Capability inventory: The agent has the ability to read project files and perform file writes when applying approved fixes. (4) Sanitization: Input code is not sanitized or filtered before being processed by the reviewer agents.

code-review