grilling

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to autonomously access the local environment. Specifically, SKILL.md states: 'When a frontier question needs a fact from the environment (filesystem, tools, etc.), dispatch a sub-agent to find it'. This grants the agent permission to perform file reads and tool execution based on its internal logic.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to drive tool-enabled sub-agents.
  • Ingestion points: User answers to the 'frontier' questions defined in SKILL.md.
  • Boundary markers: None. The prompt does not define delimiters for user input or instruct the agent to ignore instructions embedded within user answers.
  • Capability inventory: The agent is authorized to dispatch sub-agents with access to the filesystem and other tools (SKILL.md).
  • Sanitization: None. There is no mention of validating or escaping user input before it is used to determine what 'facts' the sub-agents should look up.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 03:08 AM
Security Audit — agent-trust-hub — grilling