jev-evidence-eval
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to evaluate external documents and candidate passages which may contain malicious instructions.
- Ingestion points: External text passages and documents processed via the user's harness or the suggested adapter.
- Boundary markers: None specified in the instructions; the agent is directed to treat all content as data for evaluation.
- Capability inventory: No automated tools or scripts are provided in the skill folder, but instructions imply text analysis and comparisons using external LLM providers.
- Sanitization: None described; the instructions focus on recording failures and interpretation errors including embedded instructions.
Audit Metadata