| name | review |
| description | 4-agent code review. Correctness, security, performance, style. Auto-fixes NITs. |
Activate When
/godmode:review, "review this", "check my code"
Workflow
1. Gather Diff
git diff main...HEAD
Read every changed file. >500 lines -> split into
per-directory reviews.
2. Multi-Agent Review (4 agents, parallel)
- Correctness: logic errors, off-by-ones, null,
unhandled promise rejections, missing returns
- Security: injection (SQL/XSS/cmd), auth bypass,
secrets in code, insecure defaults
- Performance: N+1 queries, O(n^2) in hot paths,
missing pagination, memory leaks
- Style: naming, dead code, inconsistent patterns,
missing types/returns
Each agent outputs:
SEVERITY | FILE:LINE | DESCRIPTION | SUGGESTED FIX
Severities: MUST-FIX, SHOULD-FIX, NIT.
IF no findings: state {category}: No issues found.
3. Merge & Deduplicate
Combine findings. Same file:line = merge, keep highest
severity, first agent's fix. Sort: MUST-FIX first.
4. Auto-Fix NITs
SAFE: imports, formatting, whitespace, comment typos.
UNSAFE (human): logic, API, security, test changes.
Each auto-fix = separate commit:
"review-autofix: {description}".
5. Verdict
Score: 0-10 (median of all agents).
- 10: 0 MUST, 0 SHOULD -> APPROVE
- 8-9: 0 MUST, 1-3 SHOULD -> APPROVE
- 5-7: 1+ MUST -> REQUEST CHANGES
- 0-4: 2+ MUST or critical security -> REJECT
IF score < 8: author responds to every MUST + SHOULD.
Re-review only modified parts. Max 3 iterations.
Quality Targets
- Target: <30min average review turnaround
