Docs

/ review

Code review: verify acceptance criteria, code quality, and test coverage.

The /review command triggers a code review. The AI performs the review; it does not submit work for someone else to review. It analyzes the code, validates against acceptance criteria, checks quality standards, and makes a pass/fail decision.

Usage

agent chat 
/review
agent chat 
review this
agent chat 
code review ENG-145

What the Review Checks

The review agent analyzes five areas:

  • Acceptance criteria: Are all checkboxes in the issue description checked? Does the code actually satisfy each criterion?
  • Code structure: Is the code well-organized, following project patterns?
  • Error handling: Are edge cases and failures handled properly?
  • Security: Are there any obvious security concerns?
  • Test coverage: Do tests exist for the new behavior?

What Happens

  1. Load context: Reads the issue, the code changes, and project standards
  2. Verify acceptance criteria: Confirms every criterion has been checked off and the code backs it up
  3. Code quality analysis: Evaluates structure, patterns, error handling, security, and performance
  4. Check test coverage: Looks for tests covering the new behavior
  5. Decision: Either approves (transitions to Testing) or requests changes (transitions back to Implementing with findings)

Important

The review agent does not write or edit code. It analyzes and documents findings. If changes are needed, it transitions the issue back to Implementing with a clear list of what needs fixing.

When to Use It

After /implement completes and the implementation agent has done its self-review pass. The issue should be in the Review state. If acceptance criteria are incomplete, the review will catch it and send the issue back.

Related Commands