claude-deep-review

Claude Deep Review

A comprehensive code review skill for Claude Code that combines architecture analysis, code review, error handling audit, type design analysis, comment verification, test coverage analysis, accessibility audit, localization review, concurrency analysis, performance analysis, code simplification, and security auditing into a single command.

Features

• 49 specialized review agents running in parallel via team-based orchestration
• File-based data flow — agent findings written to files, keeping context windows lightweight
• Dedicated synthesis — a separate agent merges all findings with a fresh context window
• Flexible scope detection — review PR changes, uncommitted work, or specific paths
• Modular aspects — run all agents or select specific review types
• Prioritized output — issues categorized as Critical, Important, or Suggestions
• Architecture health assessment — dependency mapping, cycle detection, hotspot analysis
• Graceful partial failure — if some agents fail, the report notes gaps without blocking
• Standalone — no dependencies on other plugins

Installation

Option 1: Plugin marketplace (recommended)

# Add the marketplace
/plugin marketplace add Iron-Ham/claude-deep-review

# Install the plugin
/plugin install deep-review@claude-deep-review

To update later:

/plugin marketplace update

Option 2: Clone to skills directory

# Clone to your global skills directory
git clone https://github.com/Iron-Ham/claude-deep-review.git ~/.claude/skills/deep-review

Option 3: Project-specific installation

# Clone to a specific project's skills directory
git clone https://github.com/Iron-Ham/claude-deep-review.git /path/to/your/project/.claude/skills/deep-review

Usage

Basic Usage

# Review current PR/branch changes (default)
/deep-review

# Explicit PR scope
/deep-review --pr

# Review uncommitted changes only
/deep-review --changes

# Review a specific path
/deep-review src/features

Selecting Review Aspects

# Full review with all cross-cutting agents (+ auto-detected platforms)
/deep-review full

# Code quality + error handling only
/deep-review code errors

# Type design + test coverage analysis
/deep-review types tests

# Architecture analysis only
/deep-review arch

# Performance analysis
/deep-review perf

# Code simplification suggestions
/deep-review simplify

Platform-Specific Reviews

# Explicitly include iOS reviewer
/deep-review ios --pr

# iOS + macOS reviewers
/deep-review apple --pr

# Both TypeScript frontend + backend reviewers
/deep-review ts --pr

# iOS + Android reviewers
/deep-review mobile --pr

# Next.js reviewer (Server Components, App Router)
/deep-review nextjs --pr

# Vue.js reviewer (Composition API, Nuxt)
/deep-review vue --pr

# Django reviewer (ORM, DRF, migrations)
/deep-review django --pr

# Angular reviewer (RxJS, DI, change detection)
/deep-review angular --pr

# Docker + Kubernetes reviewers
/deep-review containers --pr

# GraphQL reviewer (schema, resolvers, security)
/deep-review graphql --pr

# Terraform + Shell reviewers
/deep-review infra --pr

# Python and Rust reviewers
/deep-review python rust --pr

# Security audit (injection, auth, access control, crypto, supply chain)
/deep-review security --pr

# Agent instructions reviewer (CLAUDE.md, AGENTS.md, prompts)
/deep-review agent-instructions --pr

Platform reviewers are also automatically included when the changed files are relevant to a specific platform. For example, running /deep-review on a project with iOS Swift changes will include the iOS reviewer. The system uses its judgment to disambiguate — .swift files in a macOS project trigger the macOS reviewer (not iOS), .vue files trigger the Vue reviewer, angular.json projects trigger Angular, Dockerfile changes trigger Docker, K8s manifests trigger Kubernetes, .graphql files trigger GraphQL, .github/workflows/*.yml changes trigger GitHub Actions, CLAUDE.md/AGENTS.md/.cursorrules/.claude/ changes trigger the Agent Instructions reviewer, and .ts files in an Express app trigger the backend reviewer (not the frontend one).

Combined Examples

# Full review of PR changes (+ auto-detected platforms)
/deep-review full --pr

# Quick core review of uncommitted changes
/deep-review --changes

# Type and test analysis of specific directory
/deep-review types tests src/models

# Accessibility audit of PR
/deep-review a11y --pr

# Performance analysis of PR
/deep-review perf --pr

Review Aspects