grilling
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to gather and interpret data from the repository or external sources to resolve open design decisions. This allows for a scenario where malicious instructions hidden in those external sources could influence the agent's reasoning.
- Ingestion points: In
SKILL.md, the instructions state the agent should "Find repository or external facts yourself" and process "known facts" and "prior answers". - Boundary markers: The skill lacks instructions to use delimiters or to disregard instructions contained within the data being retrieved.
- Capability inventory: The skill relies on the agent's ability to read repository files and potentially access external web content to produce a "concise decision record".
- Sanitization: There is no logic or instruction provided to sanitize, escape, or validate the content retrieved from the environment before it is processed.
Audit Metadata