reference-hunt
Pass
Audited by Gen Agent Trust Hub on Jul 5, 2026
Risk Level: SAFENO_CODEPROMPT_INJECTION
Full Analysis
- [NO_CODE]: The skill consists entirely of markdown instructions and lacks any executable scripts or configuration files.
- [PROMPT_INJECTION]: The skill involves ingesting external source code, creating an indirect prompt injection surface.
- Ingestion points: repo paths, library names, and websites specified in user prompts.
- Boundary markers: none specified in instructions.
- Capability inventory: agent reads external data and writes new code.
- Sanitization: includes a manual checkpoint where the user must confirm a semantics summary before implementation.
Audit Metadata