The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes Git commands (diff, log) through a Python script (scripts/review_checklist.py) and direct shell instructions in SKILL.md to retrieve repository metadata and code changes for review.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted code changes from pull requests which could contain malicious instructions designed to influence the agent's review verdict.
Ingestion points: Untrusted code data enters the context via git diff commands in SKILL.md and the get_diff function in scripts/review_checklist.py.
Boundary markers: The skill uses standard Markdown code blocks to wrap diffs but lacks explicit instructions for the agent to ignore or disregard natural language commands embedded within the code being reviewed.
Capability inventory: The skill has access to Bash, Read, Grep, WebFetch, and WebSearch tools, which could be leveraged if an indirect injection is successful.
Sanitization: No content sanitization or filtering is performed on the diff output before it is presented to the agent for analysis.
[EXTERNAL_DOWNLOADS]: The skill references documentation and security guidelines from trusted and well-known sources, including OWASP and Google's Engineering Practices.

code-reviewer