fresh-eyes-grill

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted content such as staged diffs and source files, which creates a potential surface for indirect prompt injection.\n
  • Ingestion points: The skill reads user-specified files and staged diffs as part of its core logic in SKILL.md.\n
  • Boundary markers: Instructions include the use of a "bounded prompt" for subagents, which serves as a mitigation technique to limit context.\n
  • Capability inventory: The skill's workflow involves reading local files and spawning subagents for specialized analysis.\n
  • Sanitization: The skill does not explicitly detail sanitization or escaping procedures for the content of the code being reviewed before it is presented to the model.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 06:15 PM
Security Audit — agent-trust-hub — fresh-eyes-grill