parallel-code-review

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses git diff to identify changed files. This is a standard development operation used to scope the review and does not pose a security risk in this context.
  • [SAFE]: Subagents are explicitly instructed to run with readonly: true, which prevents the AI from making unauthorized modifications to the codebase during the review process.
  • [PROMPT_INJECTION]: The skill processes external code diffs which presents a surface for indirect prompt injection (Category 8). Maliciously crafted code in a PR could attempt to trick the subagent into misreporting its findings.
  • Ingestion points: File contents retrieved via git diff or PR links.
  • Boundary markers: None explicitly defined in the provided prompt templates.
  • Capability inventory: The subagents use the explore tool with readonly: true enabled.
  • Sanitization: No specific sanitization of the code diff is performed before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 06:38 PM
Security Audit — agent-trust-hub — parallel-code-review