grill-with-docs
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates by interviewing users and incorporating their feedback to settle design and architecture decisions. This ingestion of untrusted external data creates a surface for indirect prompt injection where a user might attempt to override constraints or misdirect the agent's planning logic.
- Ingestion points: User responses provided during the design interview and 'grilling' process described in SKILL.md.
- Boundary markers: The skill body instructs the agent to use the 'user's requested outcome and accepted constraints as the boundary,' which provides a logical goal-oriented constraint but lacks technical delimiters or sanitization protocols.
- Capability inventory: While the skill body refers to 'recording decisions' and 'capturing behavior in an executable issue,' no specific file-system tools, network APIs, or subprocess commands are defined in the provided YAML frontmatter or scripts.
- Sanitization: There are no explicit instructions for the agent to sanitize, validate, or filter user-provided requirements before they are used to influence the domain model or architectural trade-offs.
Audit Metadata