triage-gh-pr-comments
Pass
Audited by Gen Agent Trust Hub on Sep 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from GitHub pull request comments and review bodies.
- Ingestion points: Unresolved review threads and review bodies fetched from the GitHub MCP server (Step 2).
- Boundary markers: The instructions do not define specific structural delimiters for the comment content, but require assigning session-stable IDs (e.g., T01, T02) to maintain context isolation during evaluation.
- Capability inventory: The skill has capabilities to read local files, write to the filesystem to fix bugs, and perform git operations including commits and pushes to remote branches (Step 5).
- Sanitization: The skill contains explicit instructions to treat all ingested data as 'claims' that must be validated against evidence (files, line numbers, tests, and documentation) before any action is taken (Step 3). It also mandates user approval before any fixes are applied (Step 4).
Audit Metadata