receiving-code-review
Pass
Audited by Gen Agent Trust Hub on Jun 20, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill employs strong behavioral instructions and negative constraints (e.g., "NEVER", "Forbidden Responses") designed to override the agent's default communication patterns. It explicitly instructs the agent to disregard certain base guidelines (like expressing gratitude or agreement) to maintain a strictly technical and rigorous persona.
- [COMMAND_EXECUTION]: The instructions reference the use of standard development tools, including
grepfor codebase analysis and thegh(GitHub) CLI for interacting with pull request metadata. A template for using thegh apiendpoint to reply to review comments is provided, which is a routine task for the intended workflow. - [SAFE]: No indicators of data exfiltration, credential harvesting, or remote code execution were found. The skill's emphasis on verifying external suggestions and checking for regressions reflects a security-conscious approach to code implementation.
Audit Metadata