code-review
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes a Node.js helper script (scripts/review.mjs) that invokes the git command-line tool to identify project directories and manage review state.
- [EXTERNAL_DOWNLOADS]: The skill references supplementary documentation and engineering principles hosted on the author's GitHub repository (csark0812/toolbox).
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted code, diffs, and pull request metadata. It contains specific defensive instructions for the agent to treat this input strictly as data rather than instructions, mitigating potential prompt injection attacks from the reviewed source material.
Audit Metadata