inspect-perf
Pass
Audited by Gen Agent Trust Hub on Nov 5, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external source code provided by the user, which constitutes a surface for indirect prompt injection where malicious instructions could be embedded in code comments or data.
- Ingestion points: Source code files, directories, or subsystems identified via the mandatory 'target scope' requirement in SKILL.md.
- Boundary markers: The skill defines a 'Canonical Output Contract' to structure its reporting, but does not explicitly describe delimiters or 'ignore' instructions for the code content it reads.
- Capability inventory: The skill performs file system reads and can launch concurrent 'research' subagents for large-scale analysis (SKILL.md).
- Sanitization: No specific escaping or sanitization logic for the analyzed code is mentioned in the execution protocol.
Audit Metadata