review
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest and "run everything" from untrusted sources like Pull Requests and commits, creating a surface where malicious code or comments could attempt to influence the agent's behavior. \n
- Ingestion points: Git repositories, PR branches, and local checkouts. \n
- Boundary markers: None provided in the instructions for the data being reviewed. \n
- Capability inventory: Shell access for git operations and package installation (e.g., npm ci). \n
- Sanitization: Absent; the skill relies on the agent's internal adversarial reasoning. \n- [COMMAND_EXECUTION]: The skill directs the agent to execute shell commands, including cloning repositories and installing dependencies using tools like npm. These actions are performed on external code, which is a necessary but inherent risk for a code review tool.
Audit Metadata