rapids-ai

🌊 RAPIDS - Ship Fast, Make Money

Portable, token-optimized Claude Code framework for velocity and revenue

Install once on any machine. Get 10 AI agents, 5 MCPs, CI/CD automation, and revenue-first workflows.

RAPIDS = Rapid AI-Powered Development & Shipping Philosophy: Code without revenue is worthless. Ship fast, validate with users, iterate based on data.

⚡ Installation (One Command)

npm install -g rapids-ai

That's it! RAPIDS will automatically:

• Install 11 autonomous AI agents globally
• Configure 5 MCP servers (Context7, Filesystem, PostgreSQL, GitHub, Puppeteer)
• Set up CLI helpers (rapids-init, rapids-update)
• Display a beautiful installation progress UI

🎨 Beautiful Terminal UI

RAPIDS uses Ink by Vadim Demedes (the same framework powering Claude Code, GitHub Copilot, and Codex) for a stunning installation experience:

• 🌈 Rainbow gradient branding
• 📊 Real-time progress bars
• ✨ Animated spinners
• 🎭 BigText ASCII art
• 💫 Live MCP installation tracking

🚀 What You Get (Globally)

11 Autonomous Sub-Agents

AI agents that work in parallel (up to 3 simultaneously):

Development Agents (10):

• Feature Builder - Complete features across mobile/web/backend
• Bug Hunter - Find and fix bugs autonomously
• Code Searcher - Deep code analysis (auto-activated)
• Test Generator - Comprehensive test generation
• Refactor Master - Intelligent refactoring
• Performance Analyzer - Full-stack optimization
• DB Architect - Database schema management
• Security Scanner - Security audits
• Deployment Manager - Automated deployments
• Design Master - Material Design 3 expert for UNIQUE, brand-specific UIs (not generic templates)

💰 THE MOST CRITICAL AGENT:

• Marketing Strategist - Revenue generation, user acquisition, GTM strategy, copywriting, pricing, growth hacking

5 MCP Servers (Cutting-Edge)

Model Context Protocol - The "USB-C for AI":

• Context7 🔥 - Up-to-date documentation (auto-activated)
• Filesystem - Local file access across all projects
• PostgreSQL - Direct database querying
• GitHub - Repository management, issues, PRs
• Puppeteer - Web automation for market research

Stack Templates

• Flutter + Riverpod mobile features
• Next.js 15 App Router pages
• FastAPI + SQLAlchemy endpoints

Docker-First (Always Included)

• Dockerfiles for Next.js and FastAPI
• docker-compose.yml templates
• Multi-stage builds optimized for DokPloy

📦 Usage

Migrate Existing Project (Non-Destructive)

cd your-existing-project
rapids-migrate

Zero risk - Analyzes and adds RAPIDS without touching your code:

• Auto-detects stack (Next.js, FastAPI, Flutter)
• Identifies project structure (monorepo, multi-folder, single)
• Creates .claude/ configuration
• Initializes .agent/ documentation system
• Preserves all existing files

Read full migration guide →

Create New Project (Interactive)

rapids-init

Interactive wizard with:

• Project name input with validation
• Stack preset selection:
  • 🚀 Full Stack (Mobile + Web + Backend)
  • 📱 Mobile Only (Flutter)
  • 🌐 Web + Backend
  • ⚡ Backend Only (FastAPI)
• Automatic Dockerfile generation
• Beautiful success screen

Update RAPIDS

rapids-update

Updates to the latest version while preserving your settings.

Use in Claude Code

All 11 agents are globally available:

Launch Design Agent:

Launch design-master agent to:
- Create UNIQUE Material Design 3 color scheme with custom personality
- Design distinctive navigation with signature interactions
- Generate creative layouts that stand out while staying accessible

Launch Marketing Agent:

Launch marketing-strategist agent to:
- Create go-to-market strategy for my app
- Write landing page copy that converts
- Analyze competitors and recommend pricing

Use MCP Servers:

"use context7" to get latest Next.js 15 patterns
"use puppeteer" to scrape competitor pricing

Launch Parallel Agents:

Launch design-master + feature-builder + marketing-strategist in parallel to:
- Design UNIQUE, brand-specific UI (Material Design 3)
- Build the feature implementation
- Plan the feature launch campaign

🎯 Optimized For

• Flutter mobile apps with Riverpod
• Next.js 15 web applications
• FastAPI backends with async/await
• PostgreSQL databases
• DokPloy deployment
• Making money (Marketing & Business Strategy Agent)

📊 Productivity Gains

• Feature Development: 3-5 days → 4-8 hours (85% faster)
• Bug Fixing: 2-4 hours → 15-30 minutes (90% faster)
• Test Writing: Manual → Auto-generated
• Deployment: 30 minutes → 5 minutes (83% faster)
• Overall: ~10x productivity multiplier

🏗️ How It Works

RAPIDS installs globally to:

• macOS: ~/Library/Application Support/Claude/rapids/ → ~/.claude/
• Linux: ~/.config/claude/rapids/ → ~/.claude/

All your projects automatically have access to:

• 11 AI agents (including Material Design 3 expert)
• 5 MCP servers
• Stack templates
• Slash commands

🎨 Why RAPIDS?

Before RAPIDS:

❌ Manually configure each project
❌ Copy-paste agent configurations
❌ Inconsistent setup across projects
❌ No marketing/business strategy help

After RAPIDS:

✅ Install once, use everywhere
✅ All agents available globally
✅ Beautiful terminal experience
✅ Revenue-focused with marketing agent

📚 Documentation

• RAPIDS Method - Complete methodology
• Development Guide - Usage guide
• Sub-Agents Guide - Agent details
• Stack Config - Stack requirements

🔧 Requirements

• Node.js 20+ (for Ink TUI)
• Claude Code (latest version)
• macOS or Linux (Windows support coming soon)

🌟 Tech Stack

• Ink 5 - React for terminal UI
• React 18 - Component framework
• TypeScript - Type safety
• execa - Shell command execution
• fs-extra - Enhanced file operations

🤝 Contributing

Contributions welcome! Please read our contributing guidelines.

📄 License

MIT License - Use freely in your projects.

🙏 Credits

Built for solo developers deploying on DokPloy with:

• Anthropic Claude Code - AI pair programming
• Claude Sonnet 4.5 - Language model
• Ink by Vadim Demedes - Terminal UI framework

💰 Token Optimization (40-60% Savings)

RAPIDS achieves massive token savings through:

• Subagent Isolation - Separate context windows per agent
• Parallel Execution - Run 3 agents concurrently
• Smart Output Styles - RAPIDS Speed & Ship Mode
• Auto-compact - Trigger at 150K tokens
• Optimized Reads - Using offset, limit, head_limit

Example Savings:

• Full-stack feature: 80-120K → 35-50K tokens (~60% savings)
• Bug fix + test: 25-40K → 10-18K tokens (~55% savings)
• Marketing campaign: 30-50K → 15-25K tokens (~45% savings)

🎨 Output Styles (NEW!)

RAPIDS Speed Mode

/output-style rapids-speed

• Extremely concise, action-first
• No preamble or explanations
• Target: <50K tokens per feature

Ship Mode (Revenue-First)

/output-style ship-mode

• Every decision driven by revenue
• Validates ROI before building
• Focus on MRR, CAC, LTV, churn

🔄 CI/CD Automation (NEW!)

GitHub Actions templates included:

• @claude fix this on issues → Auto-fix
• Open PR → Auto-test + security scan
• Merge to main → Auto-deploy
• Label revenue → Marketing analysis

Setup: Add ANTHROPIC_API_KEY to GitHub Secrets, copy workflow file, done!

Version: 3.5.0 (Token-Optimized Release) Stack: Flutter • Next.js 15 • FastAPI • PostgreSQL • Docker MCPs: Context7 • Filesystem • PostgreSQL • GitHub • Puppeteer New: Output Styles • CI/CD Templates • 60% Token Savings

🌊 Install once. Ship fast. Make money. 🚀

📚 Full Documentation

• Quick Start Guide
• Token Optimization Guide
• Installation Structure
• RAPIDS Method
• Development Guide
• Subagents Guide
• Shortcuts
• Changelog

Quick Start

# Install globally (works on any machine with Claude Code)
npm install -g rapids-ai

# Run installer
rapids

# Activate revenue-first mode
/output-style ship-mode

# Build something that makes money
/agent marketing-strategist "Validate market for [idea]"
/agent feature-builder "Build [feature]"
/agent deployment-manager "Ship to production"

Now go build something that makes money. 💰