receiving-orders
Pass
Audited by Gen Agent Trust Hub on Oct 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and act upon external pull request comments and code review feedback, which represent an untrusted input surface for an AI agent.
- Ingestion points: The skill explicitly processes user-generated content from GitHub PR comments, review threads, and external reviewers ('the-inspector').
- Boundary markers: The skill lacks technical prompt delimiters (like XML tags or specific markers) but attempts to mitigate risks through a 'Response Protocol' that mandates verification against the codebase before implementation.
- Capability inventory: The skill enables the agent to perform codebase searches (using tools like 'grep'), implement code changes, and reply to specific API endpoints ('gh api').
- Sanitization: The instructions require the agent to manually verify all claims against the codebase reality, check for regressions, and test fixes individually rather than performing automated sanitization of the input text.
Audit Metadata