workflow-pr
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of GitHub PR comments and is instructed to act on that feedback.
- Ingestion points: The agent is directed in
SKILL.md(Step 5) to usegh pr view --commentsto read and analyze feedback from every thread. - Boundary markers: The instructions lack boundary markers or delimiters to help the agent distinguish between legitimate code feedback and adversarial instructions embedded in comments.
- Capability inventory: The skill utilizes significant system capabilities including executing build and test commands (
pnpm build,make build,pytest), performing git operations, and merging code viagh pr merge(SKILL.md Phase 1 and 3). - Sanitization: There is no evidence of sanitization, filtering, or validation of the comment content before it is processed by the agent to implement code fixes.
Audit Metadata