grilling

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, creating a potential surface for indirect prompt injection attacks where malicious instructions hidden in data could influence agent behavior.
  • Ingestion points: The skill is instructed to read requests, plans, repository contents, tests, documentation, references, and decisions from the user's environment (SKILL.md).
  • Boundary markers: There are no explicit instructions for the agent to use boundary markers, delimiters, or to ignore embedded instructions when processing these external sources.
  • Capability inventory: The skill utilizes filesystem search, environment exploration, and triggers specialized tools and sub-routines like /brain-dump, /plan-arbiter, /resilience-review, and /steelman (SKILL.md).
  • Sanitization: The instructions do not specify any sanitization, validation, or escaping of the content ingested from external repository files or user documents.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 02:00 AM
Security Audit — agent-trust-hub — grilling