AI-powered CLI with 24 agents for code review, debugging, testing, security audit, DevOps, API design, and more. Works with Claude, OpenAI, Copilot, Aider, Ollama.
Package Exports
dev-crew
dev-crew/dist/index.js
This package does not declare an exports field, so the exports above have been automatically detected and optimized by JSPM instead. If any package subpath is missing, it is recommended to post an issue to the original package (dev-crew) to support the "exports" field. If that is not possible, create a JSPM override to customize the exports field for this package.
Readme
Dev-Crew
24 AI Agents for Code Review, Testing, Security, DevOps & More — One CLI Works with Claude Code, OpenAI, GitHub Copilot, Aider, Ollama. Zero config. Free & open source.
Dev-Crew is a free, open-source CLI that gives you an entire AI development team from your terminal. Instead of one generic AI assistant, you get 24 specialized agents — each an expert in their domain.
npminstall-g dev-crew
dev-crew
That's it. Dev-Crew auto-detects your project stack and picks the best available AI provider. No API keys needed — simulation mode works out of the box.
Dev-Crew v2.2.0
my-app (typescript/nestjs) on main
Claude Code · 24 agents · /help for commands
❯ review @src/controllers/auth.ts
⠹ Checking git history... 4s
Smart context engine gathers files, schemas, configs, git history
Expensive token usage
Comment stripping + compression reduces tokens by 10-30%
Can't try without API keys
Simulation mode — works instantly, no setup needed
AI forgets your preferences
Feedback system — teach agents your rules, they remember across sessions
Want to use in CI/CD
CI mode — dev-crew review src/ --ci returns exit code 2 on critical issues
🚀 Quick Start
Install
# Install globally (recommended)npminstall-g dev-crew
# Or use npx (no install)
npx dev-crew
# Or install from sourcegit clone https://github.com/vasoyaprince14/dev-crew.git
cd dev-crew &&npminstall&&npm run build &&npmlink
Requirements
Node.js 18 or higher
AI provider (optional) — any one of: Claude Code, Aider, GitHub Copilot, OpenAI CLI, or Ollama
Works without any AI installed using simulation mode
First Run
# Start interactive mode (default)
dev-crew
# Or initialize in your project
dev-crew init
# Check everything works
dev-crew doctor
Your First Commands
# Review code
dev-crew review src/
# Fix a bug
dev-crew fix src/api/handler.ts --issue"null reference on line 42"# Generate tests
dev-crew test src/services/auth.ts
# Security audit
dev-crew security src/
# Ask anything
dev-crew ask "how does authentication work in this project?"
🤖 All 24 Agents
Core Agents
Agent
Command
What It Does
Reviewer
review [path]
Deep code review — knows your framework, DB, patterns
Fixer
fix <file>
Suggests fixes with diffs, can auto-apply
Debugger
debug <input>
Root cause analysis from logs, errors, stack traces
Tester
test <file>
Generates unit/integration/e2e tests for your test runner
Q&A
ask <question>
Ask anything about your codebase
Explainer
explain <file>
Detailed walkthrough of how code works
Architecture & Leadership
Agent
Command
What It Does
Tech Lead
tech-lead [question]
Architecture decisions with trade-off analysis
Business Analyst
ba <requirement>
Turns requirements into technical specs
CTO
cto review
Strategic review across 7 dimensions
PR Reviewer
pr review
Full pull request review with git diff analysis
Security
security [path]
OWASP Top 10 audit with fix guidance
Designer
designer api|schema
API and schema design review
DevOps & Infrastructure
Agent
Command
What It Does
DevOps
devops [question]
Docker, CI/CD, Terraform, Kubernetes help
Cost Optimizer
cost-optimizer
Cloud cost analysis for AWS/GCP/Azure/Vercel
Monitoring
monitoring
Observability, alerting, and logging strategy
Deploy
deploy
Complete deployment strategy for your stack
Full-Stack & Database
Agent
Command
What It Does
Scaffold
scaffold <desc>
Generate entire project structures
DB Architect
db-architect
Schema design, query optimization, indexing
API Architect
api-architect
REST/GraphQL design, versioning, rate limiting
Mobile Development
Agent
Command
What It Does
Flutter
flutter [input]
Dart/Flutter review and best practices
React Native
react-native [input]
RN performance, navigation, native modules
iOS
ios [input]
Swift/SwiftUI, memory management, App Store guidelines
Android
android [input]
Kotlin/Jetpack Compose, lifecycle, Play Store compliance
Quality & Performance
Agent
Command
What It Does
Performance
performance [path]
Frontend + backend performance audit
Accessibility
accessibility [path]
WCAG 2.1 AA compliance audit
💬 Interactive Mode
The fastest way to use Dev-Crew. Just type naturally:
dev-crew # starts interactive mode
dev-crew i # shorthand
❯ review @src/controllers/auth.ts
████████░░ 8/10
● Missing rate limiting on login endpoint
→ Add express-rate-limit middleware
✓ Good JWT token validation
✓ Password hashing with bcrypt
review · 2,341 tokens · 3.2s
❯ fix the authentication bug in @src/auth.ts
❯ write tests for the user service
❯ is there any security issue in the API?
❯ explain how the middleware chain works
❯ how much would it cost to deploy on AWS?
Slash Commands
Command
Description
/help
Show all commands and usage examples
/agents
List all 24 agents with descriptions
/clear
Clear screen
/diff
Show recent git changes
/provider <name>
Switch AI provider
/project
Show detected project info
/doctor
Check setup and providers
/tokens
Session token usage stats
/feedback <agent> <msg>
Teach an agent your preferences
/export [file]
Save last response to file
/quit
Exit (or Ctrl+C twice)
Tips
Use @path/to/file to include files in your query
Use \ at end of line for multi-line input
Tab completion works for file paths, agents, and slash commands
Ctrl+C cancels the current operation
Ctrl+C twice exits Dev-Crew
📖 Complete Guide
Code Review
# Basic review
dev-crew review src/
# Quick review (less detail, faster)
dev-crew review src/ --depth quick
# Deep review with explanations
dev-crew review src/controllers/ --depth deep --explain# Review with git history analysis
dev-crew review src/api/ --git-aware
# CI mode (JSON output, exit code 2 on critical issues)
dev-crew review src/ --ci
Fixing Code
# Auto-detect issues and suggest fixes
dev-crew fix src/services/user.ts
# Fix a specific issue
dev-crew fix src/api/handler.ts --issue"handle null response from API"# Show diff without applying
dev-crew fix src/utils/parser.ts --dry-run
# Auto-apply without confirmation
dev-crew fix src/config.ts --auto-apply
Testing
# Generate unit tests
dev-crew test src/services/auth.ts
# Integration tests
dev-crew test src/api/ --type integration
# E2E tests
dev-crew test src/routes/ --type e2e
Security Auditing
# Full security audit
dev-crew security src/
# In interactive mode
❯ check if there are any SQL injection vulnerabilities
❯ review @src/api/auth.ts for security issues
DevOps & Deployment
# Docker and CI/CD guidance
dev-crew devops "how to set up GitHub Actions for this project"# Deployment strategy
dev-crew deploy "deploy to AWS with auto-scaling"# Cost analysis
dev-crew cost-optimizer "compare AWS vs GCP for our stack"# Monitoring setup
dev-crew monitoring "set up alerting for API latency"
Database & API Design
# Schema review
dev-crew db-architect "review our user schema for performance"# API design
dev-crew api-architect "design REST endpoints for user management"
Mobile Development
# Flutter
dev-crew flutter "review state management in @lib/providers/"# React Native
dev-crew react-native "optimize list rendering performance"# iOS
dev-crew ios "review memory management in @Sources/Networking/"# Android
dev-crew android "check Jetpack Compose performance"
Multi-Agent Workflows
# Full feature pipeline: BA → Tech Lead → Security → Tests → Review
dev-crew feature "Add JWT authentication with refresh tokens"# Prioritized refactoring plan
dev-crew refactor-plan src/
# Change impact analysis
dev-crew impact src/core/auth.ts
Teaching Agents Your Preferences
# In CLI
dev-crew feedback review "Always flag console.log in production code"
dev-crew feedback security "We use Helmet.js, don't flag HTTP headers"
dev-crew feedback test"We prefer describe/it blocks over test() calls"# In interactive mode
❯ /feedback review never suggest class components, always hooks
❯ /feedback fix preserve existing comments when fixing code
Token Management
# Estimate tokens before running
dev-crew tokens estimate src/
# View usage history
dev-crew tokens usage
Dev-Crew auto-detects installed AI tools and picks the best one:
Provider
Detection
Priority
Claude Code
claude --version
1 (highest)
Aider
aider --version
2
GitHub Copilot
gh copilot
3
OpenAI CLI
openai --version
4
Ollama
ollama --version
5
Simulation
Always available
Fallback
# Check available providers
dev-crew doctor
# Switch provider in interactive mode
❯ /provider ollama
✓ Ollama
# Force simulation mode (no AI needed)
❯ /provider simulation
No providers installed? Dev-Crew runs in simulation mode — you can explore every command and see the output format without any AI backend.
# View full config
dev-crew config show
# Get a setting
dev-crew config get settings.max_tokens_per_request
# Update a setting
dev-crew config set settings.max_tokens_per_request 16000
Dev-Crew needs at least one AI tool installed. Install any of:
# Claude Code (recommended)npminstall-g @anthropic-ai/claude-code
# Or Ollama (free, runs locally)# See: https://ollama.ai# Or use simulation mode (no install needed)
dev-crew # auto-falls back to simulation
"Timeout" errors
For large codebases, target specific files:
# Instead of reviewing entire src/
dev-crew review @src/controllers/auth.ts
# In interactive mode
❯ review @src/services/user.ts
Command not found after install
# Check install locationnpm list -g dev-crew
# If using nvm, ensure correct node version
nvm use 20npminstall-g dev-crew
Contributing
Contributions welcome! Please open an issue or submit a pull request.
git clone https://github.com/vasoyaprince14/dev-crew.git
cd dev-crew
npminstallnpm run dev # Watch modenpmtest# Run testsnpm run build # Production build