engineer-review

Comprehensive multi-agent code review with parallel specialist reviewers. Use when the user says "review this", "code review", "review my PR", provides a PR number or branch, or after completing a /engineer-work cycle.

Safety Notice

This listing is imported from skills.sh public index metadata. Review upstream SKILL.md and repository scripts before running.

Copy this and send it to your AI assistant to learn

Install skill "engineer-review" with this command: npx skills add elliottrjacobs/bench-skills/elliottrjacobs-bench-skills-engineer-review

/engineer-review — Multi-Agent Code Review

Launch parallel specialist reviewer agents to comprehensively review code changes. Each reviewer focuses on one domain and reports findings independently.

When to Use

  • User says "review this", "code review", "review my PR"
  • After completing /engineer-work
  • Before merging a feature branch

Process

Step 1: Scope the Review

Determine what to review:

  • If a PR number is provided: gh pr diff [number]
  • If on a feature branch: git diff main...HEAD
  • If $ARGUMENTS specifies files: review those files

Get the diff and list of changed files.

Step 2: Detect Tech Stack

Read project config files to determine which conditional reviewers to launch:

  • tsconfig.json → TypeScript reviewer
  • next.config.* → Next.js reviewer
  • app.json or expo in package.json → Expo/RN reviewer
  • supabase/ directory → Supabase reviewer
  • docs/prds/ or docs/tech-specs/ → Spec compliance reviewer

Step 3: Launch Parallel Reviewers

Spawn ALL selected reviewers IN PARALLEL using the Task tool. Send all Task calls in a single message.

Always launch these 4 core reviewers:

Security Reviewer

prompt: Review these code changes for security vulnerabilities.
  Focus on: auth bypass, injection (SQL/XSS/command), data exposure,
  hardcoded secrets, insecure defaults, missing input validation.
  Return: file, line, severity (P1-P4), description, fix suggestion.

Performance Reviewer

prompt: Review these code changes for performance issues.
  Focus on: N+1 queries, missing indexes, unnecessary re-renders,
  bundle size impact, request waterfalls, missing caching, large payloads.
  Return: file, line, severity (P1-P4), description, fix suggestion.

Architecture Reviewer

prompt: Review these code changes for architectural issues.
  Focus on: component boundaries, module dependencies, state management
  choices, data flow patterns, separation of concerns, SOLID principles.
  Return: file, line, severity (P1-P4), description, fix suggestion.

Patterns Reviewer

prompt: Review these code changes for consistency with codebase patterns.
  First read existing files to understand conventions, then check:
  naming conventions, import patterns, error handling patterns,
  duplication, anti-patterns, TypeScript usage.
  Return: file, line, severity (P1-P4), description, fix suggestion.

Conditionally launch based on Step 2 detection:

TypeScript Reviewer (if tsconfig.json)

prompt: Review for TypeScript quality. Focus on: type safety, proper generics,
  Zod schema integration, avoiding any/as assertions, discriminated unions.

Next.js Reviewer (if next.config.*)

prompt: Review for Next.js App Router best practices. Focus on: server/client
  component boundaries, data fetching patterns, caching, middleware, server actions.

Expo/RN Reviewer (if app.json/expo)

prompt: Review for Expo/React Native patterns. Focus on: Expo Router conventions,
  NativeWind styling, platform handling, native module usage, mobile performance.

Supabase Reviewer (if supabase/ directory)

prompt: Review Supabase usage. Focus on: RLS policy completeness, auth patterns,
  client selection (browser vs server), storage policies, realtime security.

Spec Compliance Reviewer (if docs/prds/ or docs/tech-specs/)

prompt: Compare implementation against the spec. Read the latest PRD/tech-spec
  in docs/. Check: requirements met, deviations justified, nothing over-built.

See references/reviewer-catalog.md for full reviewer focus areas.

Step 4: Synthesize Results

Collect all reviewer findings and produce a unified summary:

## Code Review Summary

### P1 — Critical (must fix before merge)
| # | File | Line | Issue | Reviewer |
|---|------|------|-------|----------|

### P2 — Important (should fix)
| # | File | Line | Issue | Reviewer |

### P3 — Suggestion (consider fixing)
| # | File | Line | Issue | Reviewer |

### P4 — Nitpick (optional)
| # | File | Line | Issue | Reviewer |

### Positive Patterns
[Good patterns worth noting]

### Reviewers Run
[List which reviewers were launched and why]

Deduplicate findings across reviewers. If two reviewers flag the same issue, keep the more specific one.

Output

Review summary presented inline. Optionally save to docs/reviews/ if requested.

Next Steps

  • Issues found? Fix them and re-run /engineer-review
  • All clear? Merge the PR
  • Want to capture learnings? → /knowledge-compound

Source Transparency

This detail page is rendered from real SKILL.md content. Trust labels are metadata-based hints, not a safety guarantee.

Related Skills

Related by shared tags or category signals.

General

product-tech-spec

No summary provided by upstream source.

Repository SourceNeeds Review
General

product-prd

No summary provided by upstream source.

Repository SourceNeeds Review
Research

knowledge-compound

No summary provided by upstream source.

Repository SourceNeeds Review
General

product-naming

No summary provided by upstream source.

Repository SourceNeeds Review