proof-assistant
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data which could potentially contain hidden instructions.
- Ingestion points: Untrusted data enters the agent context through the
algorithm,code,claim, andpartialProoffields defined in the input schema ofSKILL.md. - Boundary markers: The instructions do not specify explicit delimiters or "ignore previous instructions" markers to isolate user-provided code and claims from the agent's logic.
- Capability inventory: The skill is configured with
Read,Write,Grep, andGlobtools in itsallowed-toolsfrontmatter, allowing it to interact with the file system. - Sanitization: No input validation or sanitization mechanisms are defined to filter or escape instructions embedded within the provided algorithm or code snippets.
Audit Metadata