The Agent Skills Directory

[SAFE]: Standard functional instructions for code review. The skill uses git and gh to analyze local changes and pull requests, which is consistent with its stated purpose. No malicious overrides or bypasses are present in the instructions.
[PROMPT_INJECTION]: Indirect prompt injection surface (Category 8). The skill processes untrusted code changes and pull request diffs which enters the agent's context during the review process (Step 1). The prompts for sub-agents (Step 3) lack explicit boundary markers or instructions to disregard embedded commands in the reviewed content. The skill possesses the capability to modify project files based on these analyses (Step 5). However, the risk is mitigated by a mandatory user confirmation step (AskUserQuestion) before any changes are executed, ensuring human oversight over the output.

code-review