code-reviewer

Pass

Audited by Gen Agent Trust Hub on May 29, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes untrusted code which could contain instructions intended to influence the agent.
  • Ingestion points: Processes code snippets and diffs as described in the 'Overview' and 'Flow' sections of SKILL.md.
  • Boundary markers: No specific boundary markers or instructions to ignore embedded commands are included in the prompt.
  • Capability inventory: The skill defined in SKILL.md lacks access to execution tools or system commands, which mitigates the risk of injection.
  • Sanitization: No input validation or filtering is defined for the analyzed code.
Audit Metadata
Risk Level
SAFE
Analyzed
May 29, 2026, 04:57 PM
Security Audit — agent-trust-hub — code-reviewer