grilling
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to autonomously gather facts from the local environment (filesystem and tools) via sub-agents to resolve questions related to user-supplied plans. This behavior creates a vulnerability surface where malicious instructions embedded in the user input or in the files being queried could manipulate the sub-agent's behavior.\n
- Ingestion points: The skill ingests user-provided plans, decisions, and ideas, as well as iterative answers provided throughout the 'grilling' process (SKILL.md).\n
- Boundary markers: The instructions lack delimiters or explicit directives to the agent to treat user-supplied data or environment facts as potentially untrusted instructions.\n
- Capability inventory: The skill provides the agent with the ability to dispatch sub-agents to access the local filesystem and other environment tools to resolve 'facts' (SKILL.md).\n
- Sanitization: No sanitization, validation, or human-in-the-loop review is specified for the data retrieved from the environment before the agent acts upon it.
Audit Metadata