workflow-review-changes

Pass

Audited by Gen Agent Trust Hub on Nov 4, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is primarily designed to review pull requests and external code changes, which is a significant ingestion point for untrusted data. Maliciously crafted code or comments in the reviewed content could attempt to influence the agent's logic or override the review workflow.
  • Ingestion points: Processes pull request changes, local worktree files, and merge-base contributions (SKILL.md).
  • Boundary markers: The skill does not define explicit boundary markers or "ignore previous instructions" delimiters for the code content being analyzed.
  • Capability inventory: The skill is authorized to spawn sub-agents for specialized reviews, write report files to the tmp/ directory, and execute local Node.js scripts (SKILL.md, references/fix-loop.md).
  • Sanitization: There are no explicit instructions for sanitizing or escaping the content of the files under review before they are processed by the agent or sub-agents.
  • [COMMAND_EXECUTION]: The skill lifecycle involves executing local Node.js scripts within the project's .claude/ directory to synchronize contracts and manage review snapshots.
  • Evidence: The instructions direct the agent to run node .claude/scripts/lib/workflow-skill-contract.cjs --write for registry synchronization and node .claude/hooks/lib/review-receipt.cjs for capturing and issuing review snapshots (SKILL.md, references/fix-loop.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 4, 2026, 10:12 AM