📖 FrootAI Setup Guide

Two tools, one setup page. Get FrootAI’s MCP Server (for your AI agent) and VS Code Extension (for you) up and running in minutes.

📦 Part 1: MCP Server Setup

Add AI architecture knowledge to any agent. Works with Claude Desktop, VS Code Copilot, Cursor, Windsurf, Azure MS Foundry, and any MCP-compatible client.

Prerequisites

Node.js 18+ — Download here
Git — Download here
An MCP client — VS Code, Claude Desktop, Cursor, Windsurf, or Azure AI Foundry

Step 1: Install

Option A: npm (Recommended — zero clone)

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

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

Published on npmjs.com/package/frootai-mcp — 664KB knowledge base bundled inside. Works from anywhere.

Option B: From GitHub (for development)

# Clone the repository
git clone https://github.com/gitpavleenbali/frootai.git

# Navigate to MCP server
cd frootai/mcp-server

# Install dependencies
npm install

That's it — the MCP server is ready. The knowledge.json (664 KB) contains all 18 modules bundled. No additional setup needed.

Step 2: Connect to Your Client

Choose your client and add the FrootAI server to its MCP configuration:

Claude Desktop / Cursor / Windsurf

Open your config file and add the FrootAI server:

Config file location:
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Cursor: Settings → MCP Servers → Add

{
  "mcpServers": {
    "frootai": {
      "command": "node",
      "args": ["/absolute/path/to/frootai/mcp-server/index.js"]
    }
  }
}

✅ Restart Claude Desktop / Cursor. FrootAI will appear in your tools list.

Step 3: Verify It Works

Ask your agent any of these to confirm FrootAI is connected:

"What is temperature in LLMs?"

→ lookup_term

Returns precise glossary definition

"List all FrootAI modules"

→ list_modules

Shows 5 FROOT layers with 18 modules

"How do I design a RAG pipeline?"

→ get_architecture_pattern

Returns complete RAG design guide

"Search for hallucination reduction"

→ search_knowledge

Returns top sections across modules

Available Tools

Tool	Parameters	Returns
list_modules	none	All 18 modules organized by FROOT layer
get_module	module_id (F1–T3), section? (optional)	Full module content or specific section
lookup_term	term (string)	Glossary definition from 200+ terms
search_knowledge	query (string), max_results? (1–10)	Ranked matching sections across modules
get_architecture_pattern	scenario (enum)	Pre-built decision guide with recommendations
get_froot_overview	none	Complete FROOT framework summary
fetch_azure_docs	service (string)	Live Azure docs from Microsoft Learn (falls back to static)
fetch_external_mcp	query (string)	MCP servers from registries (falls back to curated list)
list_community_plays	filter? (string)	20 solution plays from GitHub (falls back to bundled list)
get_github_agentic_os	primitive? (enum)	.github agentic OS guide per primitive

Architecture Pattern Scenarios

rag_pipeline
RAG Pipeline

agent_hosting
Agent Hosting

model_selection
Model Selection

cost_optimization
Cost Optimization

deterministic_ai
Deterministic AI

multi_agent
Multi-Agent

fine_tuning_decision
Fine-Tuning Decision

💻 Part 2: VS Code Extension Setup

Browse solution plays, search 200+ AI terms, initialize DevKit — right from your editor sidebar.

Step 1: Install

Option A (Marketplace): Open VS Code → Extensions (Ctrl+Shift+X) → Search "FrootAI" → Install

Option B (Terminal):

code --install-extension pavleenbali.frootai

✅ Reload VS Code. FrootAI panels appear in your sidebar.

Step 2: Explore the Sidebar

After install, you'll see 3 panels in the sidebar:

📋

Solution Plays

Browse all 20 plays. Click to open README or folder.

📖

FROOT Modules

18 modules grouped by FROOT layer.

🔌

MCP Tools

See all 13 MCP tools at a glance.

Step 3: Use Commands (Ctrl+Shift+P)

FrootAI: Look Up AI TermType any AI/ML term → jumps to definition

FrootAI: Search Knowledge BaseFull-text search across all 18 docs

FrootAI: Initialize DevKitFull .github Agentic OS (19 files) + agent.md + MCP + plugin.json

FrootAI: Initialize HooksCopy guardrails.json (preToolUse policy gates)

FrootAI: Initialize PromptsCopy 4 slash commands (/deploy, /test, /review, /evaluate)

FrootAI: Open Solution PlayOpens a play's README with option to open folder

FrootAI: Show Architecture PatternPick from 7 patterns: RAG, agents, hosting, cost

FrootAI: Open Setup GuideOpens this setup guide page

FrootAI: Browse Solution PlaysOpens solution plays page on the website

Step 4: Initialize DevKit for a Solution Play

1. Run Ctrl+Shift+P → FrootAI: Initialize DevKit

2. Select a solution play (e.g., Enterprise RAG)

3. FrootAI copies the full .github Agentic OS (19 files) + agent.md + .vscode/mcp.json + plugin.json to your workspace

4. Start coding — Copilot has full agentic context: instructions, prompts, agents, skills, hooks, workflows.

Troubleshooting

Extension not showing in sidebar

Reload VS Code (Ctrl+Shift+P → 'Reload Window'). Ensure v0.2.0+ is installed.

Commands not appearing

Check that FrootAI is enabled in Extensions. Try reinstalling: code --install-extension pavleenbali.frootai

DevKit Init copies nothing

Ensure you have a workspace folder open. FrootAI copies files to the root of your active workspace.

MCP tools panel empty

This is informational — it shows what tools the MCP server exposes. Install the MCP server separately (see Part 1).

General Troubleshooting

Server not appearing in tool list

Restart your MCP client after adding the config. Ensure the path to index.js is absolute.

Module not found

Run `npm run build` in mcp-server/ to regenerate knowledge.json from the docs/ folder.

Node.js version error

FrootAI requires Node.js 18+. Check with `node --version`.

Permission denied on macOS/Linux

Run `chmod +x mcp-server/index.js` to make the entry point executable.

🔗 Back to Ecosystem 📦 MCP Server 💻 VS Code Extension 🌳 Back to FrootAI