code-review

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill contains only natural language instructions and configuration; it lacks executable scripts, network commands, or system-level modifications.
  • [SAFE]: No obfuscation, prompt injection, or persistence patterns were found.
  • [INDIRECT_PROMPT_INJECTION]: The skill reviews untrusted code, which is an inherent injection surface. The instructions explicitly direct the agent to identify security risks in that code, which serves as a defensive measure. Evidence: 1. Ingestion points: Diffs and files via git status (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: No dangerous primitives listed. 4. Sanitization: Absent.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 12:00 PM
Security Audit — agent-trust-hub — code-review