grilling

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill uses behavioral instructions to define a specific persona ("grilling") for stress-testing user plans. These instructions guide conversational flow and do not attempt to bypass safety guidelines or override agent constraints.
  • [COMMAND_EXECUTION]: The instructions permit the agent to explore the environment (filesystem and tools) to retrieve facts. This is an intended capability for the agent to gather necessary context for its analysis.
  • [DATA_EXPOSURE]: While the skill allows the agent to read facts from the environment, there are no instructions for external data transmission, exfiltration, or hardcoded credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 10:24 AM
Security Audit — agent-trust-hub — grilling