The Agent Skills Directory

[SAFE]: The skill defines a strict, non-prose output format for code reviews, which helps maintain consistency and prevents conversational drift during automated tasks.
[SAFE]: It leverages standard developer tools like git and gh (GitHub CLI) to access the code context required for the review process.
[SAFE]: The instructions explicitly direct the agent to prioritize security vulnerabilities such as injections, authorization bypasses, and exposed secrets in the code it reviews.
[PROMPT_INJECTION]: The skill processes untrusted external data (source code from files, diffs, and pull requests), which presents a surface for indirect prompt injection where malicious instructions could be embedded in the code being reviewed. This is a standard risk for any review-based skill and is mitigated by the agent's core safety protocols.

signal-review