bento-eval-blind
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill directs the agent to dynamically load and follow instructions from repository-specific files like
AGENTS.mdand.agents/bento/bento-eval-blind.md. This ingestion of external instructions without sanitization or boundary markers creates a surface for indirect prompt injection. - Ingestion points:
AGENTS.md,CLAUDE.md, and.agents/bento/bento-eval-blind.mdin the project root. - Boundary markers: Absent.
- Capability inventory: Agent tools such as file system access and shell execution are available to instructions defined in the ingested files.
- Sanitization: Absent.
Audit Metadata