grilling
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied ideas and environmental data (filesystem, tools) to build a branching 'design tree', which creates a surface where untrusted data could influence agent behavior.
- Ingestion points: User responses to rounds of questions in SKILL.md.
- Boundary markers: The skill instructions do not specify any delimiters or warnings to ignore instructions embedded within the user's answers or tool outputs.
- Capability inventory: The skill explicitly allows for tools to access the filesystem and other utilities via sub-agents as described in SKILL.md.
- Sanitization: There is no evidence of sanitization or validation of the facts retrieved from the environment or the user's responses before they are used to compute the decision tree.
Audit Metadata