jmb-grilling
Pass
Audited by Gen Agent Trust Hub on Nov 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to automatically 'find facts' from the environment, specifically referencing the filesystem and other tools, via sub-agents. This introduces a vulnerability surface where the agent might process untrusted data that contains hidden instructions.
- Ingestion points: Untrusted data enters the agent context through filesystem reads and tool outputs during 'frontier' rounds (SKILL.md).
- Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to ignore instructions embedded in the retrieved data.
- Capability inventory: The skill utilizes sub-agent dispatching and general tool access to explore the environment.
- Sanitization: Absent. There is no mention of sanitizing, escaping, or validating the content retrieved from the environment before it is used to inform the 'design tree'.
Audit Metadata