grilling
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill implements a logic-based 'design tree' approach to user interaction. It guides the agent through iterative rounds of questioning to reach a shared understanding.
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for the agent to dispatch sub-agents to fetch 'facts from the environment (filesystem, tools, etc.)'. This constitutes a standard capability surface for AI agents.
- Ingestion points: Processes user-provided plans and ideas as input.
- Boundary markers: No explicit delimiters are specified for user-provided context or sub-agent responses.
- Capability inventory: File system access and tool execution via sub-agents (SKILL.md).
- Sanitization: No specific sanitization or instruction-guarding is implemented for data retrieved from the environment.
Audit Metadata