testing-policy-creator
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted codebase data to generate a new testing policy skill.\n
- Ingestion points: The 'Survey the codebase' step in SKILL.md involves reading file structures, exported surfaces, and existing test assertions across the repository.\n
- Boundary markers: No explicit delimiters or instruction-bypass warnings are defined for the survey phase where conclusion data is interpolated into the final SKILL.md template.\n
- Capability inventory: The skill possesses file-writing capabilities (.claude/skills/ directory, AGENTS.md, and system temp files) and utilizes the Skill tool for issue creation and interactive 'grilling'.\n
- Sanitization: No explicit sanitization or validation of the surveyed code data is mentioned before it is included in the policy output.\n- [SAFE]: The skill's primary function is codebase analysis and documentation. It follows established software engineering frameworks (e.g., Khorikov, Google SWE) and incorporates human-in-the-loop validation via the 'Grill' step before committing any changes to the repository.
Audit Metadata