grill-me

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill provides a logical framework for plan analysis and adversarial interviewing. No malicious logic, unauthorized network operations, or hidden payloads were identified.
  • [INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it processes user-supplied plans.
  • Ingestion points: User-provided plans or ideas for the agent to 'grill' in SKILL.md.
  • Boundary markers: The skill uses a structured round-based questioning format and design tree mapping to organize output.
  • Capability inventory: The skill utilizes search_files, read_file, terminal, and delegate_task across its instructions.
  • Sanitization: The instructions do not specify sanitization for user-provided input, relying on the agent's logic to differentiate facts from user claims.
  • [COMMAND_EXECUTION]: The skill instructs the agent to use the terminal tool to retrieve environmental facts, which is an intended function of the planning phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 02:31 PM
Security Audit — agent-trust-hub — grill-me