audit-fix
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes project source code that could contain embedded instructions designed to manipulate the audit results or influence the fixes applied by the agent.
- Ingestion points: Project files located in the current working directory or specified path.
- Boundary markers: The skill includes a 'PROVENANCE NOTE' to provide context to the auditor, though it lacks formal structural delimiters around the code being analyzed.
- Capability inventory: The skill possesses file modification capabilities and the ability to execute local test suites (e.g., npm, pytest, go test, cargo test).
- Sanitization: The skill relies on the model's internal safety protocols and specific instructions to perform the audit with 'full rigor' rather than explicit input sanitization.
Audit Metadata