senior-python
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill documentation consists entirely of high-quality engineering guidelines and reference materials. It contains no executable scripts, remote downloads, or credential exposures. The instructions promote secure coding practices and rigorous verification.\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze potentially untrusted repository code. While this represents a data ingestion surface, the instructions include significant safeguards, advising the agent to treat external inputs as evidence rather than proof, verify claims against official documentation, and maintain high suspicion of AI-generated content. Evidence Chain:\n
- Ingestion points: SKILL.md (target or diff, direct callers and callees, nearby tests, manifests, lockfiles)\n
- Boundary markers: Present (SKILL.md: "Separate facts from assumptions", "State unknowns instead of filling them with assumptions")\n
- Capability inventory: Limited to analysis tools (review, diagnosis, design, refactoring)\n
- Sanitization: Present (references/security-and-trust.md: "Parse untrusted data once near the boundary... reject or safely normalize ambiguity")
Audit Metadata