JSPM

  • Created
  • Published
  • Downloads 63
  • Score
    100M100P100Q102770F
  • License MIT

Build complete apps from a single prompt. AI-powered CLI with 25 agents — app builder, code review, testing, security, DevOps. Works on top of Claude Code, free, no API keys.

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

npm version monthly downloads total downloads GitHub stars license node version

Dev-Crew

Build Complete Apps from a Single Prompt — 25 AI Agents, One CLI
App builder + code review, testing, security, DevOps. Works on Claude Code. Free & open source.

App BuilderQuick StartAll AgentsInteractive ModeGuideProviders


What is Dev-Crew?

Dev-Crew is a free, open-source CLI that builds complete apps from a single prompt — and gives you 25 specialized AI agents for every stage of development.

npm install -g dev-crew

# Build a complete app
dev-crew create "build an uber clone with React and Node.js"

# Or use any of the 25 agents
dev-crew review @src/app.ts
dev-crew test @src/auth.ts
dev-crew security @src/

No API keys needed. Works on top of Claude Code (free). Zero config.


Build an App in One Command

dev-crew create "build a todo app with Next.js and PostgreSQL"

Dev-Crew asks you a few questions, then runs a 6-stage AI pipeline:

  Step 1: Requirements    — AI business analyst scopes your app
  Step 2: Architecture    — Tech lead designs the system
  Step 3: Code Generation — App creator writes every file
  Step 4: Database & API  — DB architect refines schema
  Step 5: Tests           — Test agent generates test suite
  Step 6: DevOps          — Docker, CI/CD, deployment configs

Result: A complete, runnable project written to disk with npm install and git init done automatically.

# Options
dev-crew create "SaaS dashboard" --stack "nextjs+postgres" --yes
dev-crew create "chat app" --output ./my-chat-app --no-install

Why Dev-Crew Over Other AI Tools?

Problem Dev-Crew's Answer
Want to build a full app but don't know where to start dev-crew create — describe your app, get production-ready code
Generic AI gives surface-level reviews 25 domain-expert agents — security agent knows OWASP, DevOps agent knows Kubernetes
Need to copy-paste code into ChatGPT Auto-detects your framework, DB, ORM and builds context automatically
Locked to one AI provider 5 providers — Claude, OpenAI, Copilot, Aider, Ollama + simulation fallback
AI doesn't know your codebase 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 modedev-crew review src/ --ci returns exit code 2 on critical issues

🚀 Quick Start

Install

# Install globally (recommended)
npm install -g dev-crew

# Or use npx (no install)
npx dev-crew

# Or install from source
git clone https://github.com/vasoyaprince14/dev-crew.git
cd dev-crew && npm install && npm run build && npm link

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 25 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 25 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

CI/CD Integration

# .github/workflows/review.yml
name: Code Review
on: [pull_request]
jobs:
  review:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: 20
      - run: npm install -g dev-crew
      - run: dev-crew review src/ --ci
        # Exit code 0 = pass, 2 = critical issues found

🔌 Multi-Provider Support

Dev-Crew works on top of existing AI tools — no extra API keys needed. Just install Claude Code (or any provider) and Dev-Crew uses it automatically.

Provider Detection Priority Notes
Claude Code claude --version 1 (default) No extra setup — just works
Aider aider --version 2
GitHub Copilot gh copilot 3
OpenAI CLI openai --version 4
Ollama ollama --version 5 Free, runs locally
Claude API ANTHROPIC_API_KEY Optional Direct API — real streaming, exact tokens
Simulation Always available Fallback Works without any AI installed

Default Setup (Zero Config)

# If you have Claude Code installed, Dev-Crew just works
npm install -g dev-crew
dev-crew

Optional: Direct Claude API

For power users who want real token-by-token streaming and exact token counts:

# Optional — only if you want direct API access
export ANTHROPIC_API_KEY=sk-ant-...
npm install @anthropic-ai/sdk

# Then switch in interactive mode
❯ /provider claude-api

Switching Providers

# Check available providers
dev-crew doctor

# Switch in interactive mode
❯ /provider claude-code    # Default
❯ /provider ollama         # Local, free
❯ /provider simulation     # No AI needed

No providers installed? Dev-Crew runs in simulation mode — explore every command without any AI backend.


⚙️ Configuration

Initialize Project Config

dev-crew init

Creates .dev-crew/config.yml:

project:
  name: my-app
  stack: nestjs
  database: postgresql
  orm: prisma
  test_framework: vitest

settings:
  max_tokens_per_request: 8000
  show_token_usage: true
  auto_include_schema: true
  confirm_before_apply: true

agents:
  review:
    severity: normal
    rules:
      - "Always check for SQL injection in raw queries"
      - "Enforce strict TypeScript — no 'any' types"
    ignore:
      - "**/*.spec.ts"
      - "src/generated/**"
    focus:
      - security
      - performance
      - error-handling
  test:
    framework: vitest
    coverage_target: 80

Configuration Commands

# 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

🛠️ Supported Stacks

Dev-Crew auto-detects your technology stack:

Category Supported
Languages TypeScript, JavaScript, Python, Go, Rust, Java, Kotlin, Swift, Dart, C#, Ruby, PHP, C/C++
Web Frameworks NestJS, Express, Fastify, Hono, Next.js, Nuxt, React, Vue, Angular, Svelte, Astro, Django, Flask, FastAPI, Rails, Laravel, Spring Boot
Mobile Flutter, React Native, SwiftUI/UIKit, Jetpack Compose
Databases PostgreSQL, MySQL, MongoDB, Redis, SQLite
ORMs Prisma, Drizzle, TypeORM, Sequelize, Mongoose, Knex
Testing Vitest, Jest, Mocha, Pytest, Go testing
CI/CD GitHub Actions, GitLab CI, Jenkins, CircleCI
Package Managers npm, pnpm, yarn, bun

📋 All 48 Commands

Click to expand full command list

Setup

Command Description
dev-crew init Initialize Dev-Crew in your project
dev-crew doctor Check system setup and requirements

Core

Command Description
dev-crew review [path] Code review (--depth, --explain, --git-aware, --ci)
dev-crew fix <file> Fix issues (--issue, --auto-apply, --dry-run)
dev-crew debug <input> Debug from logs/errors/stack traces
dev-crew test <file> Generate tests (--type unit|integration|e2e)

Advanced

Command Description
dev-crew tech-lead [question] Architecture decisions
dev-crew ba <requirement> Requirements to specs
dev-crew cto <action> Strategic technical review
dev-crew pr <action> PR review (--branch)
dev-crew security [path] Security audit
dev-crew designer <type> API/schema design review

Smart Features

Command Description
dev-crew ask <question> Codebase Q&A
dev-crew explain <file> Code explanation
dev-crew onboard New developer guide
dev-crew watch [path] Auto-review on file changes
dev-crew impact <file> Change impact analysis
dev-crew resolve Merge conflict resolution
dev-crew interactive Interactive REPL mode

Multi-Agent

Command Description
dev-crew feature <desc> BA → Tech Lead → Security → Tests → Review
dev-crew refactor-plan <path> Prioritized refactoring plan

Code Health

Command Description
dev-crew debt <action> Technical debt tracker
dev-crew migration-check <path> Migration safety check
dev-crew deps <action> Dependency health scanner

DevOps

Command Description
dev-crew devops [question] Docker, CI/CD, infrastructure
dev-crew cost-optimizer [question] Cost analysis
dev-crew deploy [question] Deployment strategy
dev-crew monitoring [question] Observability and alerting

Full-Stack

Command Description
dev-crew create <desc> Build complete app from a prompt
dev-crew scaffold <desc> Scaffold new project
dev-crew build <desc> Build a feature (--stack)
dev-crew fullstack-builder <desc> Full-stack architecture
dev-crew db-architect [input] Database design
dev-crew api-architect [input] API design

Mobile

Command Description
dev-crew flutter [input] Flutter/Dart
dev-crew react-native [input] React Native
dev-crew ios [input] iOS/Swift
dev-crew android [input] Android/Kotlin

Quality

Command Description
dev-crew performance [path] Performance audit
dev-crew accessibility [path] WCAG compliance

Analytics

Command Description
dev-crew tokens <action> Token management
dev-crew analytics Developer analytics
dev-crew sprint-report Sprint summary (--days)
dev-crew stats npm download stats

Configuration

Command Description
dev-crew agents <action> List/inspect agents
dev-crew config <action> Manage config
dev-crew feedback <agent> <msg> Teach agents
dev-crew patterns View learned patterns

Architecture

dev-crew
├── bin/dev-crew.ts              # CLI entry point (Commander.js)
├── src/
│   ├── agents/                  # 24 specialized AI agents
│   │   ├── base-agent.ts        # Shared execution engine
│   │   ├── registry.ts          # Agent factory
│   │   └── */agent.ts           # Individual agents
│   ├── commands/                # CLI command handlers
│   │   └── interactive.ts       # REPL mode
│   ├── core/
│   │   ├── provider-bridge.ts   # Multi-provider abstraction
│   │   ├── nlp-router.ts        # Natural language → agent routing
│   │   ├── context-engine.ts    # Smart context gathering
│   │   ├── token-optimizer.ts   # Token compression
│   │   ├── project-detector.ts  # Stack auto-detection
│   │   └── config-manager.ts    # YAML config management
│   ├── features/                # Debt tracker, patterns, analytics
│   └── types/                   # TypeScript interfaces
├── .github/workflows/           # CI + npm stats
└── tsup.config.ts               # Build config (ESM, Node 18+)

Dev-Crew vs Other Tools

Feature Dev-Crew Claude Code GitHub Copilot Cursor
Specialized agents (24)
Code review with framework context Limited
Security audit (OWASP) Manual
Test generation Manual Limited
DevOps/Docker/K8s guidance Manual
Mobile (Flutter/RN/iOS/Android)
DB schema design
API architecture review
Multi-provider (5 backends) Claude only Copilot only Multiple
Works without API key
Feedback/learning system
Interactive REPL
CI/CD integration
Open source
Price Free Paid Paid Paid

Troubleshooting

"No AI provider found"

Dev-Crew needs at least one AI tool installed. Install any of:

# Claude Code (recommended)
npm install -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 location
npm list -g dev-crew

# If using nvm, ensure correct node version
nvm use 20
npm install -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
npm install
npm run dev     # Watch mode
npm test        # Run tests
npm run build   # Production build

License

MIT License — see LICENSE for details.


Built by Prince Vasoya
npmReport BugRequest Feature

If Dev-Crew helps you, please give it a ⭐ on GitHub — it helps others discover it!