grilling
Pass
Audited by Gen Agent Trust Hub on Aug 9, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The instructions direct the agent to autonomously retrieve 'facts' from the filesystem and environment tools based on user-provided input. This interaction model presents an indirect prompt injection surface where the agent might encounter and act upon malicious instructions contained within files it reads while attempting to verify user claims.\n
- Ingestion points: User-provided plans, decisions, or ideas entered during the 'grilling' session in
SKILL.md.\n - Boundary markers: None. The skill does not define delimiters for external data or instruct the agent to ignore instructions embedded in retrieved facts.\n
- Capability inventory: The skill explicitly encourages the use of sub-agents to explore the 'environment (filesystem, tools, etc.)' in
SKILL.md.\n - Sanitization: None. There are no instructions to sanitize or validate the content retrieved from the environment before processing.
Audit Metadata