merge-risk
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes Pull Request data, which is untrusted content from external authors. This represents an attack surface where a malicious actor could attempt to influence the agent's risk assessment via instructions embedded in the PR description or code comments. However, the skill lacks the capabilities (such as network access or command execution) to perform harmful actions based on such an injection.
- Ingestion points: Pull Request descriptions, code diffs, and comments analyzed during execution.
- Boundary markers: None specified.
- Capability inventory: The skill is purely analytical and does not invoke shell commands, file operations, or network requests.
- Sanitization: No sanitization is performed on the ingested PR text.
Audit Metadata