Skip to Content
Getting StartedQuick Start

Quick Start

Go from zero to a working FrootAI environment in under 5 minutes.

Prerequisites

  • Node.js 18+ — nodejs.org  (22+ recommended)
  • Git — git-scm.com 
  • An AI-capable editor: VS Code with GitHub Copilot, Cursor, or Claude Desktop

Step 1 — Install FrootAI

# MCP Server — add AI tooling to any MCP-compatible agent
npx frootai-mcp@latest
 
# SDK — programmatic API for Node.js / TypeScript
npm install frootai

Step 2 — Configure Your AI Agent

Add FrootAI as an MCP server in your editor’s config:

mcp.json (VS Code / Claude Desktop / Cursor)
{
  "mcpServers": {
    "frootai": {
      "command": "npx",
      "args": ["frootai-mcp@latest"]
    }
  }
}
💡

This works with GitHub Copilot, Claude Desktop, Cursor, Windsurf, Azure AI Foundry, and any MCP-compatible client.

Step 3 — Init a Solution Play

Scaffold a complete DevKit for Enterprise RAG (Play 01):

npx frootai init-devkit 01

This generates the full play structure:

solution-plays/01-enterprise-rag/
├── .github/                    # DevKit: Copilot brain
│   ├── copilot-instructions.md #   Always-on context
│   ├── agents/                 #   builder, reviewer, tuner
│   ├── instructions/           #   Coding standards
│   ├── skills/                 #   Step-by-step procedures
│   └── hooks/                  #   Security guardrails
├── config/                     # TuneKit: AI parameters
│   ├── openai.json             #   Model + temperature
│   └── guardrails.json         #   Quality thresholds
├── infra/                      # Azure Bicep templates
├── spec/                       # Architecture decisions
└── fai-manifest.json           # FAI Protocol wiring

Step 4 — Run Your First Evaluation

Validate that all primitives are correctly wired:

# Validate schemas, naming, and frontmatter
npm run validate:primitives
 
# Load the play in the FAI Engine
node engine/index.js solution-plays/01-enterprise-rag/fai-manifest.json --status

Expected output:

🍊 FAI Engine v0.1
══════════════════════════════════════════════════
  Play:      01-enterprise-rag v1.0.0
  Knowledge: 2 FROOT modules
  WAF:       3 pillars enforced
  Guardrails: groundedness≥0.95, safety≥0.99
  ✅ All primitives wired, context resolved
══════════════════════════════════════════════════

Step 5 — Explore with the SDK

import { FrootAI } from 'frootai';
 
const fai = new FrootAI();
 
// Search across all knowledge modules
const results = fai.search('RAG architecture');
 
// Get play details
const play = fai.plays.get('01');
console.log(play.title);  // "Enterprise RAG Q&A"
 
// Run evaluation quality gates
fai.evaluation.run({
  groundedness: 0.95,
  relevance: 0.88,
  safety: 1.0
});
ℹ️

The SDK works offline with zero external dependencies — all knowledge is bundled.

What’s Next?

Last updated on
InstallationGuides