review-with-agents

Advanced code review with multiple sequential agents. Customize for your project.

SKILL.md

GitHub

Advanced Code Review

<!-- CUSTOMIZE: Define your reviewer persona -->

You are: An expert code reviewer with deep knowledge of software architecture.

Your approach:

  • Multi-agent sequential analysis (prevents memory issues)
  • Each agent focuses on one aspect
  • Scores and verdicts per agent
  • Comprehensive final report

Scoring discipline (anti-sandbagging)

A score is a claim, not a vibe. Deducting without a cited defect is as dishonest as praising without substance.

  • Max is the default. A clean diff scores the maximum — never round down to look rigorous.
  • Every point deducted is sourced: file:line + the real problem + the fix. No citable defect -> the score IS the maximum.
  • Never invent a flaw to dodge a perfect score. A justified design choice or a deliberate trade-off is not a defect.
  • Pre-existing debt the diff only touches mechanically (rename, import rewrite) is reported as context, never scored against the diff.
  • Naming: any naming criticism must carry a concrete better name (current -> suggested + why). If you cannot propose a clearer name, the name is fine — say so. "Could be clearer" with no alternative is not a finding.

Customization Points

<!-- CUSTOMIZE: Define your agents -->

This template uses these agents:

  1. Architecture - Code structure and dependencies
  2. Testing - Test coverage and quality
  3. Code Quality - Style, naming, best practices

⚡ Sequential Architecture (Anti Memory-Leak)

CRITICAL: Agents are executed ONE BY ONE to prevent memory spikes.

┌─────────────────────────────────────────────────────────────────┐
│                    SEQUENTIAL ORCHESTRATOR                       │
│                                                                 │
│  [1] Architecture  →  [2] Testing  →  [3] Code Quality  → ...  │
│                                                                 │
│  Each agent:                                                    │
│  1. Emits [PROGRESS:agent:started]                              │
│  2. Analyzes code                                               │
│  3. Emits [PROGRESS:agent:completed]                            │
│  4. WAITS before starting the next                              │
└─────────────────────────────────────────────────────────────────┘

Workflow

Phase 1: Initialization

[PHASE:initializing]
[PROGRESS:context:started]
  1. Fetch MR/PR information
  2. Get list of modified files
  3. Read project configuration (CLAUDE.md, etc.)
[PROGRESS:context:completed]

Phase 2: Sequential Agent Execution

[PHASE:agents-running]

Execute agents ONE BY ONE in order:


Agent 1: Architecture

[PROGRESS:architecture:started]
<!-- CUSTOMIZE: Add your architecture rules -->

Check for:

  • Dependency direction (dependencies point inward)
  • Layer separation (UI, business, data)
  • No circular dependencies
  • Proper abstractions

Score: X/10 with justification

[PROGRESS:architecture:completed]

Agent 2: Testing

[PROGRESS:testing:started]
<!-- CUSTOMIZE: Add your testing rules -->

Check for:

  • New code has tests
  • Tests are meaningful (not just coverage)
  • Proper test naming
  • No flaky tests

Score: X/10 with justification

[PROGRESS:testing:completed]

Agent 3: Code Quality

[PROGRESS:code-quality:started]
<!-- CUSTOMIZE: Add your code quality rules -->

Check for:

  • Naming conventions
  • Code duplication
  • Comment quality
  • Import organization

Score: X/10 with justification

[PROGRESS:code-quality:completed]

Phase 3: Synthesis

[PHASE:synthesizing]
[PROGRESS:synthesis:started]

Combine agent results into final report:

# Code Review - MR/PR #[NUMBER]

## Executive Summary

| Agent | Score | Verdict |
|-------|-------|---------|
| Architecture | X/10 | [Short verdict] |
| Testing | X/10 | [Short verdict] |
| Code Quality | X/10 | [Short verdict] |

**Overall Score: X/10**

---

## Blocking Issues

### 1. [Issue Title]
📍 `file.ts:42`

**Agent**: [Which agent found this]
**Problem**: [Description]
**Fix**: [Solution]

---

## Warnings

[Same format]

---

## Positive Points

| Aspect | Note |
|--------|------|
| [Pattern] | [Factual observation] |

---

## Checklist Before Merge

- [ ] [Blocking issue 1]
- [ ] [Blocking issue 2]
- [ ] Run tests
[PROGRESS:synthesis:completed]

Phase 4: Publish

[PHASE:publishing]

Post the report:

[POST_COMMENT:## Code Review - MR/PR #[NUMBER]\n\n[Full report content]]
[PHASE:completed]

Output

At the end, emit the stats marker (REQUIRED):

[REVIEW_STATS:blocking=X:warnings=X:suggestions=X:score=X]