Start Review Command

PRIMARY OBJECTIVE

Professional code review using specialized agents. Ensure quality, architectural compliance, and requirements fulfillment before merge.

WORKFLOW

GATE 1: Task Identification

1. AskUserQuestion: "Which task to review? Provide task path or PR URL."

2. Validate:

   • Task document exists with "Implementation Complete" status
   • STOP if: "In Progress" or missing PR information
   • Linear issue referenced, steps marked complete

GATE 2: Automated Quality Gate

ACTION: Invoke automated-quality-gate agent

Task directory: [path]
Run quality checks (format, lint, types, tests, build) and return structured results.
Agent returns data inline (no file created) - integrate into Code Review.

• STOP if GATE_FAILED → Return to developer with fixes list
• Agent returns JSON + markdown snippet for integration into Code Review

GATE 3: Approach Review

ACTION: Invoke senior-architecture-reviewer agent

Task directory: [path]
Review approach, requirements fulfillment, architecture fit, TDD compliance (git history).

• STOP if NEEDS_REWORK → Return to developer with issues

GATE 4: Parallel Code Review

ACTION: Invoke these 5 agents IN PARALLEL:

1. security-code-reviewer:

   Copy

   Review security: OWASP, input validation, auth. Task: [path]
   

2. code-quality-reviewer:

   Copy

   Review quality: SOLID, DRY, patterns, DDD layers. Task: [path]
   

3. test-coverage-reviewer:

   Copy

   Review tests: coverage gaps, edge cases, quality. Task: [path]
   

4. documentation-accuracy-reviewer:

   Copy

   Review docs: accuracy, completeness. Task: [path]
   

5. performance-reviewer:

   Copy

   Review performance: bottlenecks, N+1 queries, efficiency. Task: [path]
   

GATE 5: Synthesis & Decision

TEMPLATE: Use @docs/product-docs/templates/code-review-template.md

1. Read template from docs/product-docs/templates/code-review-template.md
2. Aggregate findings from all agents:
   • Quality Gate (GATE 2)
   • Approach Review (GATE 3)
   • All 5 Code Review agents (GATE 4)
3. Apply severity markers [CRITICAL], [MAJOR], [MINOR], [INFO] within each agent section
4. Consolidate issues into severity categories with source agent attribution
5. Apply decision matrix (see below)
6. Create Code Review - [Task].md in task directory using template structure

GATE 6: Codex Final Review (If APPROVED)

CONDITION: Only run if Decision Matrix result is APPROVED (0 Critical, 0-2 Major)

ACTION: Invoke /codex-cli skill for cross-AI final validation

Verification focus: Final pre-merge review as last reviewer before main

• All acceptance criteria implemented
• No regressions or breaking changes
• Code is production-ready
• No security issues missed by previous reviewers

Process feedback:

• STOP if CRITICAL issues → Change decision to NEEDS FIXES
• Add findings to Code Review document under "Cross-AI Validation" section

GATE 7: Linear & Completion

1. Update Linear status (end only) using the cc-linear skill:

   • Do this as a single atomic operation (status update only).
   • Target status: [Ready to Merge | Needs Fixes | In Review]

2. Add results comment (end only) using the cc-linear skill:

   • Do this as a separate atomic operation (comment only).
   • Include status + issue counts + review doc path.

3. Notify user of outcome and next steps

DECISION MATRIX

Severity Levels:

• [CRITICAL] - Must fix before merge (blocks approval)
• [MAJOR] - Should fix (3+ blocks approval)
• [MINOR] - Nice to fix (does not block)
• [INFO] - Observations (does not block)

Status Mapping:

• APPROVED → "Ready to Merge"
• NEEDS FIXES → "Needs Fixes"
• NEEDS DISCUSSION → Keep "In Review"

OUTPUT

Single Code Review - [Task].md created in task directory using template from @docs/product-docs/templates/code-review-template.md.

Important: All review data is consolidated into this single file. No separate files are created for:

• Quality Gate Report (integrated into Pre-Review Validation section)
• Approach Review (integrated into Pre-Review Validation section)

Contains:

• Pre-Review validation (Quality Gate + Approach Review inline)
• Code Review findings from 5 specialized agents (with inline severity markers)
• Cross-AI Validation (Codex final review - if APPROVED)
• Consolidated issues checklist with agent attribution
• Decision with severity counts and next steps