grilling

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to autonomously fetch environmental facts from the filesystem and tools whenever a question requires them, rather than asking the user. This automated ingestion of data creates a surface for processing untrusted content that might contain hidden instructions.
  • Ingestion points: Filesystem and environmental tool outputs (SKILL.md).
  • Boundary markers: Absent. The instructions do not define delimiters or provide guidance to the agent to ignore potentially malicious embedded instructions in the fetched data.
  • Capability inventory: Sub-agent dispatch, filesystem access, and generic tool usage (SKILL.md).
  • Sanitization: Absent. No validation, escaping, or filtering of retrieved content is specified before the data is integrated into the decision tree.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 12:36 PM
Security Audit — agent-trust-hub — grilling