kata-review-pr
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from pull requests which can contain malicious instructions to the agent. Ingestion points: PR metadata, comments, and changed files (SKILL.md). Boundary markers: None explicitly mentioned to separate PR content from instructions. Capability inventory: Capability to run local tests and use CLI tools for repository inspection (SKILL.md). Sanitization: No sanitization of external content is described.
- [COMMAND_EXECUTION]: The skill instructions allow the agent to run shell commands for repository inspection and testing. Evidence: 'Use the available code-host tool or CLI' and 'Run focused tests only when review authority and the local checkout make that safe' in SKILL.md.
Audit Metadata