Installation

FrootAI is available through multiple distribution channels. Choose the one that fits your workflow.

Prerequisites

Requirement	Minimum Version	Check
Node.js	18+ (22+ recommended)	node --version
Python	3.9+ (for Python SDK)	python --version
Docker	20+ (for container usage)	docker --version
Git	2.30+	git --version
Azure CLI	2.60+ (for deployment)	az version

info

Only Node.js is required for the core tooling. Python, Docker, and Azure CLI are needed only for their respective channels.

npm — MCP Server

The MCP Server exposes 45 tools to any AI agent that speaks the Model Context Protocol:

# Run directly (no install needed)
npx frootai-mcp@latest

# Or install globally
npm install -g frootai-mcp
frootai-mcp

Add to your editor's MCP configuration:

.vscode/mcp.json

{
  "mcpServers": {
    "frootai": {
      "command": "npx",
      "args": ["frootai-mcp@latest"]
    }
  }
}

tip

This configuration works with GitHub Copilot, Claude Desktop, Cursor, Windsurf, and any MCP client.

npm — SDK & CLI

The SDK provides a full programmatic API for Node.js and TypeScript:

npm install frootai

import { FrootAI } from 'frootai';

const fai = new FrootAI();
fai.search('RAG architecture');   // BM25-ranked results
fai.plays.get('01');              // Play details
fai.listModules();                // 16 FROOT modules

The CLI provides command-line access to all FrootAI features:

# Run via npx (no install)
npx frootai --help

# Or install globally
npm install -g frootai
frootai init-devkit 01
frootai primitives --type agents

Python — SDK

The Python SDK is offline-capable with zero external dependencies:

pip install frootai

from frootai import FrootAI

fai = FrootAI()
results = fai.search("RAG architecture")
play = fai.plays.get("01")

Python — MCP Server

pip install frootai-mcp

claude_desktop_config.json

{
  "mcpServers": {
    "frootai": {
      "command": "frootai-mcp"
    }
  }
}

VS Code Extension

Install from the Visual Studio Marketplace:

code --install-extension frootai.frootai-vscode

Or search "FrootAI" in the VS Code Extensions panel (Ctrl+Shift+X).

The extension provides:

• Solution play browsing and scaffolding
• Primitive validation and linting
• Integrated MCP server management
• DevKit initialization commands

Docker

Run the MCP server with zero local dependencies:

# Pull and run
docker pull ghcr.io/frootai/frootai-mcp
docker run -i ghcr.io/frootai/frootai-mcp

# Or use docker compose

docker-compose.yml

services:
  frootai-mcp:
    image: ghcr.io/frootai/frootai-mcp:latest
    stdin_open: true

Clone the Repository

For contributing or exploring the full source:

git clone https://github.com/frootai/frootai.git
cd frootai
npm install
npm run validate:primitives  # Verify everything works

warning

The repository includes 100 solution plays, 238 agents, 322 skills, and 77 plugins. The initial clone is approximately 50 MB.

Distribution Channels Summary

Channel	Install Command	Use Case
npm MCP	npx frootai-mcp@latest	Add FrootAI tools to any AI agent
npm SDK	npm install frootai	Programmatic API in Node.js/TS
PyPI SDK	pip install frootai	Python API, offline, zero deps
PyPI MCP	pip install frootai-mcp	Python MCP server
VS Code	code --install-extension frootai.frootai-vscode	IDE integration
Docker	docker run -i ghcr.io/frootai/frootai-mcp	Zero-install MCP server
CLI	npx frootai	Command-line access
GitHub	git clone	Full source, contributions

Next Steps

• Quick Start — 5-minute guide from zero to running
• Your First Solution Play — build and deploy Play 01