Skip to main content

What it does

/review runs a structured code review against your implementation plan, requirements, and quality standards. It checks plan compliance, code quality, security, and performance — and assigns severity levels to each issue found. This is Step 4 in the core workflow — run after /implement and before /git.

When to use

Use /review after implementation is complete to validate the work before committing. Also useful as a standalone review of any code changes.

Prerequisites

  • Implementation complete
  • Tests passing
  • An implementation plan to review against (for plan compliance checks)

Conversation mode

Either mode works.

What happens

1

Load the code-reviewer agent

The code-reviewer agent is loaded for the full review process.
2

Plan compliance check

Does the implementation match what was planned? Are all acceptance criteria met? Is there any scope creep — things added that weren’t in the plan?
3

Code quality check

Is the code readable without comments to explain it? Are functions doing one thing? Is error handling present at system boundaries? Are there tests for the new behavior?
4

Security check

Any user input that isn’t validated? Any secrets or credentials in the code? Any new endpoints missing authorization?
5

Performance check

Any N+1 queries (loop containing a database call)? Any synchronous blocking I/O in an async context? Any unbounded data fetching (missing LIMIT on queries)? Any unnecessary re-computation inside loops?
6

Report findings with severity

Each issue is reported with a severity level: Critical (blocks shipping), High (should fix before shipping), Medium (fix soon), or Low (nice to have).

Skills invoked

  • code-reviewer agent — full review process across all four checklist domains

Issue severity

SeverityMeaning
CriticalBlocks shipping — security vulnerabilities, broken functionality
HighShould fix before shipping — significant quality issues
MediumFix soon — technical debt, missing tests
LowNice to have — style, minor improvements

Example

/review
Antigravity outputs:
=== CODE REVIEW ===
Task reviewed: notification delivery implementation
Reviewer: code-reviewer agent

PLAN COMPLIANCE: PASS
- All acceptance criteria met
- No scope creep detected

CODE QUALITY: 2 issues
- MEDIUM: service.ts:87 — function is 94 lines, extract retry logic to separate function
- LOW: Missing error message context in catch block at service.ts:124

SECURITY: PASS (no auth/input handling changes)

PERFORMANCE: 1 issue
- HIGH: service.ts:52 — N+1 query inside notification loop, batch the lookups

OVERALL: CHANGES REQUESTED
Blocker count: 1 | Suggestions: 2

/implement

Step 3 — /review runs after implementation is complete.

/git

Step 5 — commit and push after /review passes.

/analyze

For deeper technical analysis beyond what /review’s checklist covers.

/improve

Apply the improvements that /review identifies.

Build docs developers (and LLMs) love