Review Loop

---
description: Run automated parallel review loop with all reviewers until CRITICAL/HIGH issues are resolved
model: sonnet
---

You are coordinating an automated review and fix loop. Your role is to orchestrate reviewers and specialists to iteratively improve code quality.

## Overview

1. Launch reviewers in parallel (background): 3 built-in reviewers + CodeRabbit CLI (optional)
2. Collect their findings
3. If CRITICAL/HIGH issues exist → Fix them with specialists
4. Re-review after fixes
5. Repeat until no CRITICAL/HIGH issues remain (max 5 iterations)

## Iteration Loop

For each iteration (max 5):

### Step 1: Launch Reviewers in Parallel

Launch all reviewers with `run_in_background=true` in a **single message**.
Always launch the 3 built-in reviewers. Additionally launch coderabbit-reviewer as a 4th reviewer (it will self-skip if CLI is not installed).

**test-reviewer:**

Review test quality and coverage, including detection of missing tests for production code changes.

1. Run git diff main...HEAD --name-only to identify changed production code files (exclude test files).
2. For each changed production file, verify that corresponding tests exist and cover the changes.
3. Review existing test files for quality, coverage, and anti-patterns.

Report ALL issues found with their severity levels (CRITICAL, HIGH, MEDIUM, LOW).

Missing tests for new or significantly changed production code should be reported as HIGH severity.

For CRITICAL and HIGH severity issues, provide:

• Severity level
• File path and line number
• Clear description of the problem
• Specific recommendation for fixing

Format your findings clearly so they can be easily identified and addressed.

**code-quality-reviewer:**

Review production code for design quality, maintainability, and adherence to project patterns.

Report ALL issues found with their severity levels (CRITICAL, HIGH, MEDIUM, LOW).

For CRITICAL and HIGH severity issues, provide:

• Severity level
• File path and line number
• Clear description of the problem
• Specific recommendation for fixing

Format your findings clearly so they can be easily identified and addressed.

**ux-architecture-reviewer:**

Review for UX architecture issues: state consistency, WebSocket/REST API contracts, session/worker lifecycle handling, and edge cases.

Report ALL issues found with their severity levels (CRITICAL, HIGH, MEDIUM, LOW).

For CRITICAL and HIGH severity issues, provide:

• Severity level
• File path and line number
• Clear description of the problem
• Specific recommendation for fixing

Format your findings clearly so they can be easily identified and addressed.

**coderabbit-reviewer (optional - self-skips if CLI not installed):**
```text
Run CodeRabbit CLI review against the main branch.

Translate all findings into standard severity format (CRITICAL, HIGH, MEDIUM, LOW).

For CRITICAL and HIGH severity issues, provide:
- Severity level
- File path and line number
- Clear description of the problem
- Specific recommendation for fixing

Format your findings clearly so they can be easily identified and addressed.

Step 2: Collect Results

Use TaskOutput with block=true to wait for each reviewer to complete:

test_result = TaskOutput(test_reviewer_id, block=true)
quality_result = TaskOutput(quality_reviewer_id, block=true)
ux_result = TaskOutput(ux_reviewer_id, block=true)
coderabbit_result = TaskOutput(coderabbit_reviewer_id, block=true)

Note: If coderabbit-reviewer reports "Skipped" (CLI not installed), simply ignore its results and continue with the other reviewers' findings.

Step 3: Extract CRITICAL/HIGH Issues

From each reviewer's output, identify all CRITICAL and HIGH severity issues.

Group issues by component:

• packages/client/** → frontend issues
• packages/server/** → backend issues
• packages/shared/** → classify based on primary consumer

Step 4: Fix Issues

If CRITICAL/HIGH issues exist:

For frontend issues: Launch frontend-specialist with the list of issues to fix:

Fix the following CRITICAL/HIGH issues in packages/client:

[List all frontend issues with file, line, description, and recommendation]

After fixing each issue:
1. Ensure the fix addresses the root cause
2. Update related tests if needed
3. Verify code still follows frontend standards

After all fixes, run: bun run test

For backend issues: Launch backend-specialist with the list of issues to fix:

Fix the following CRITICAL/HIGH issues in packages/server:

[List all backend issues with file, line, description, and recommendation]

After fixing each issue:
1. Ensure the fix addresses the root cause
2. Update related tests if needed
3. Verify code still follows backend standards

After all fixes, run: bun run test

Note: If you have both frontend and backend issues, launch both specialists in parallel in a single message.

Step 5: Verify Fixes

After specialists complete, verify:

1. All CRITICAL/HIGH issues were addressed
2. Tests pass (bun run test)
3. No new issues were introduced

Step 6: Decide Next Action

• If CRITICAL/HIGH issues remain: Start next iteration
• If no CRITICAL/HIGH issues: Proceed to Step 7 (Code Simplification)
• If max iterations (5) reached: Exit loop and report remaining issues

Step 7: Code Simplification

After all CRITICAL/HIGH issues are resolved, run code-simplifier to refine the modified code:

Launch code-simplifier:

Review and simplify all files that were modified during this review-loop session.

Focus on:
- Clarity and consistency improvements
- Removing unnecessary complexity introduced during fixes
- Applying project coding standards

Preserve all functionality - only improve how the code is written.

After simplification, run: bun run test && bun run typecheck

Wait for completion. If tests or typecheck fail, report the issue but do not iterate further.

Final Report

After the loop completes, provide a s ```