verify-implementation-contracts
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [NO_CODE]: The skill consists solely of instructions and goals in markdown format, containing no executable scripts or software dependencies.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external implementation code and specifications, creating a surface for indirect instructions. (1) Ingestion points: Authoritative specification and changed implementation boundary files mentioned in SKILL.md. (2) Boundary markers: None identified. (3) Capability inventory: Analysis via parsing and reading code with no network or file-write permissions. (4) Sanitization: None identified.
- [SAFE]: The skill instructions focus on read-only code analysis and reporting, with no requests for elevated privileges, persistence mechanisms, or obfuscated code.
Audit Metadata