Work in Progress These docs are a work in progress. More content is coming soon.

toolforest.io Documentation

Getting Started

Introduction

toolforest.io is a consumer-focused MCP (Model Context Protocol) aggregator that makes it simple to connect your AI assistants and LLMs to the apps and APIs you use every day.

Instead of configuring multiple MCP servers and managing different connections, toolforest.io provides a single, unified endpoint where you can mix and match the exact tools you need. Think of it as your personal toolkit forestβ€”a curated collection of tools that grows with your needs.

Key Features

  • Single Endpoint - One MCP connection for all your tools
  • Mix & Match - Choose only the tools you need from a growing collection
  • Secure Authentication - Industry-standard OAuth integration with major platforms
  • Privacy First - Your data stays yours, no permanent storage
  • LLM Optimized - Built specifically for AI assistants like Claude, ChatGPT, and Le Chat

Quick Start

Get started with toolforest.io in just a few minutes:

1. Create an Account

Sign up at www.toolforest.io using your email account.

2. Login and Create an MCP Server Endpoint

Create an MCP server endpoint specific to your LLM client (e.g., Claude or ChatGPT).

3. Configure your toolkits.

Choose which toolkits you want to use and, where relevant, connect them to your external accounts (e.g., Google Sheets or Fitbit).

4. Connect Your LLM to toolforest.io

Add the toolforest.io MCP endpoint to your AI assistant's configuration and connect via OAuth 2.0. You can also connect your agent directly to toolforest.io via MCP with an API Key.

βœ“ You're Ready!
Your AI assistant can now access your connected tools through toolforest.io. Try asking it to read a spreadsheet or create a slide deck!

Core Concepts

MCP Overview

The Model Context Protocol (MCP) is an open protocol that standardizes how AI assistants connect to external tools and data sources. Think of it as a universal adapter that lets your LLM talk to different apps and services.

toolforest.io implements MCP to provide a seamless bridge between your AI assistant and the tools you use every day. When you connect to toolforest.io, your LLM gains access to all your authorized tools through a single, secure connection.

How It Works

  1. Your LLM connects to the toolforest.io MCP server
  2. You authorize access to specific services (Google, Microsoft, etc.)
  3. Your LLM discovers tools available through your connections
  4. You interact naturally - ask your LLM to perform tasks using your tools
  5. toolforest.io handles authentication, API calls, and data formatting

Authentication

toolforest.io uses industry-standard OAuth 2.0 authentication to connect to third-party services. This means:

  • You control exactly what permissions you grant
  • You can revoke access at any time
  • We never see your passwords
  • Tokens are encrypted and stored securely

Connecting Your AI Assistants

πŸ“‹ Prerequisites
Before you can connect your LLM, you will need to:

ChatGPT requires a paid plan for custom MCP connectors. If you don't have one, you could try Claude or Le Chat instead β€” both support MCP connectors on their free plans.

ChatGPT

Available on: Web, Mobile (iOS/Android)

Follow these steps to add the Toolforest MCP Connector to ChatGPT and enable your ChatGPT assistant to access your Toolforest tools.

Setup Steps

  1. Go to https://chatgpt.com/#settings/Connectors/Advanced

  2. Toggle Developer mode ON if it isn't already.

  3. Click Back to return to the Apps screen.

  4. Click Create (or New connector) to start adding a new connector.

  5. In Name, enter: Toolforest

  6. (Optional) In Description, enter something like: Connects ChatGPT to Toolforest

  7. In MCP Server URL, enter: https://mcp.toolforest.io/mcp

  8. In Authentication, confirm that OAuth is selected (it should be the default).

  9. Review the custom server warning and check the box: "I understand and want to continue."

    Beta Notice: This warning appears because Toolforest is currently in beta and custom MCP servers are unverified by OpenAI. We're working toward formal verification to streamline this process.

  10. Click Create. You'll see a brief loading screen while ChatGPT establishes a connection with Toolforest.

  11. If you aren't already logged into the toolforest.io portal, you'll be redirected to the Toolforest login screen. Sign in with your Toolforest account.

  12. After logging in, Toolforest will ask you to grant ChatGPT permission to access your tools. Click Approve.

  13. After a few seconds, you'll be redirected back to the ChatGPT Apps screen. You should see a confirmation that Toolforest has been connected.

  14. You can now use your Toolforest-powered tools directly from ChatGPT.

Mobile (iOS / Android)

Connecting Toolforest via the web automatically makes your tools available in the ChatGPT mobile app. No additional setup is needed.

Adding Tools to a Chat

Follow the steps below to make your toolkits available to a chat.

Steps

  1. Click on the + icon on the bottom left part of the chat window.

  2. Select Toolforest from the menu.

  3. A Toolforest indicator should appear underneath the chat window, indicating that Toolforest is connected to the chat.

Claude

Available on: Web, Mobile (iOS/Android), Desktop (Cowork), Claude Code

Follow these steps to add the Toolforest MCP Connector to Claude.

Setup Steps

  1. Open the Connectors settings:

  2. Click the + icon at the top of the screen to bring up the Add Custom Connector dialog.

  3. In the Name field, enter: Toolforest

  4. In the Remote MCP server URL field, enter: https://mcp.toolforest.io/mcp

  5. Click Add to save the configuration.

  6. Select Toolforest from the list of connectors and click the Connect button that appears on the right side of the screen.

  7. You will be redirected to the Toolforest authorization page. Click Approve.

  8. You will be redirected back to Claude settings with a success message. You should now see your connected tools on the screen.

Mobile (iOS / Android)

Connecting Toolforest via the Web or Desktop App (Cowork) automatically makes your tools available in the Claude mobile app. No additional setup is needed.

Claude Code (Terminal and Web)

Connecting Toolforest via the Web or Desktop App (Cowork) automatically makes your tools available in Claude Code via MCP. No additional setup is needed.

Le Chat

Available on: Web, Mobile (iOS/Android)

Follow these steps to connect Toolforest to Mistral's Le Chat interface.

Setup Steps

  1. Go to https://chat.mistral.ai/connections

  2. Click the + Add Connector button in the top right corner.

  3. In the popup window, switch to the Custom MCP Connector tab.

  4. In Connector Name, enter: Toolforest

  5. In Connector Server, enter: https://mcp.toolforest.io/mcp

  6. (Optional) Add a description.

  7. In Authentication Method, select OAuth2.1 if it's not already selected.

  8. Click Connect.

  9. You will be redirected to the Toolforest authorization page. Click Approve.

  10. Once redirected back to Le Chat, scroll down to the Connectors list to confirm that Toolforest is active (indicated by a green status light).

Mobile (iOS / Android)

Connecting Toolforest via the web automatically makes your tools available in the Le Chat mobile app. No additional setup is needed.

Adding Tools to a Chat

Follow the steps below to make your toolkits available to a chat.

Steps

  1. Click on the + icon on the bottom left part of the chat window.

  2. Select Toolforest from the menu.

  3. A Toolforest indicator should appear underneath the chat window, indicating that Toolforest is connected to the chat.

OpenClaw

Available on: CLI / Self-hosted agents

Follow these steps to connect your OpenClaw agents to Toolforest and give them access to your configured toolkits.

OpenClaw dashboard showing API Keys and IP Allowlist sections

Setup Steps

  1. Log in to toolforest.io and select your OpenClaw MCP server from the sidebar.

  2. Click the Connect tab in the top right corner.

  3. Under IP Allowlist, enter the source IP address of your OpenClaw instance (the machine where your agent runs) and click Add. You can find your current IP by running curl ifconfig.me. If your machine uses IPv6, add that address as well.

  4. Click + Generate New API Key. Give the key a descriptive name (e.g. "my-openclaw-agent") and click Generate.

  5. Copy the key immediately β€” it will only be shown once. The key starts with tfo_.

  6. Verify your key works by listing the available tools:

curl -X POST https://mcp.toolforest.io/mcp \
  -H "Authorization: Bearer tfo_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'

You should see a JSON response listing your enabled toolkits. If you get a 401 or 403 error, double-check your API key and IP allowlist.

Connect to OpenClaw

Method 1: CLI Install (Recommended)

Use this if you have terminal access to the machine running OpenClaw.

Step 1 β€” Install the plugin

openclaw plugins install @toolforest/toolforest-plugin@latest

Step 2 β€” Set your API key

Use the API key you generated in the steps above.

openclaw config set plugins.entries.toolforest.config.apiKey "tfo_your_key_here"

Step 3 β€” Make tools available on all profiles

Warning: Do not use openclaw config set tools.alsoAllow '["toolforest"]' β€” that replaces the entire list and removes any other plugins you have configured. Use the append command below instead.

existing=$(openclaw config get tools.alsoAllow 2>/dev/null || echo '[]')
openclaw config set tools.alsoAllow "$(node -e "const v=JSON.parse('$existing'); if(!v.includes('toolforest'))v.push('toolforest'); console.log(JSON.stringify(v))")"

Step 4 β€” Restart the gateway

openclaw gateway restart

Note: In some environments the restart process gets killed when the gateway shuts down. If the agent becomes unresponsive after this step, restart the gateway manually (e.g. openclaw gateway start).

Step 5 β€” Verify

openclaw plugins inspect toolforest

You should see: toolforest: Registered 4 meta-tools

Method 2: Agent-Assisted Install

Use this if you prefer to let your OpenClaw agent handle the setup. Results may vary depending on your LLM.

Send this message to your OpenClaw agent:

Set up the Toolforest plugin for OpenClaw. My API key is: tfo_your_key_here.
Read the setup instructions from https://registry.npmjs.org/@toolforest/toolforest-plugin
and follow them exactly.

Replace tfo_your_key_here with the API key you generated in the steps above.

The agent will install the plugin, configure your API key, and restart the gateway. If it gets stuck, use Method 1.

Troubleshooting

  • Check that your API key is correct and active. You can regenerate it from the dashboard if needed.
  • Verify your current IP matches an entry in the IP Allowlist section. If you're behind a VPN or proxy, add that IP as well. Run curl ifconfig.me to check your current IP.
  • Enable the required service (e.g. Gmail, Google Calendar) in your Toolforest dashboard under Toolkits.
  • If the gateway becomes unresponsive after installation, try restarting it manually with openclaw gateway start.

Generic LLM Agent

Available on: CLI / Self-hosted agents

Follow these steps to connect your custom LLM agent or application to Toolforest and give it access to your configured toolkits.

Toolforest portal showing API Keys and IP Allowlist sections

Setup Steps

  1. Log in to toolforest.io and select your MCP server from the sidebar, or create a new one for your agent.

  2. Click the Connect tab in the top right corner.

  3. Under IP Allowlist, enter the source IP address of the machine where your agent runs and click Add. You can find your current IP by running curl ifconfig.me. If your machine uses IPv6, add that address as well.

  4. Click + Generate New API Key. Give the key a descriptive name (e.g. "my-langchain-agent") and click Generate.

  5. Copy the key immediately β€” it will only be shown once. The key starts with tfo_.

  6. Verify your key works by listing the available tools:

curl -X POST https://mcp.toolforest.io/mcp \
  -H "Authorization: Bearer tfo_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","method":"tools/list","id":1}'

You should see a JSON response listing your enabled toolkits. If you get a 401 or 403 error, double-check your API key and IP allowlist.

Connect to Your Agent

Any MCP-compatible framework can connect to Toolforest by pointing at https://mcp.toolforest.io/mcp with your API key as a bearer token.

Step 1 β€” Install the adapter

pip install langchain-mcp-adapters langgraph

Step 2 β€” Connect and load tools

# Connect any LangChain / LangGraph agent to Toolforest
from langchain_mcp_adapters.client import MultiServerMCPClient
from langchain.agents import create_agent

client = MultiServerMCPClient({
    "toolforest": {
        "url": "https://mcp.toolforest.io/mcp",
        "transport": "http",
        "headers": {"Authorization": "Bearer tfo_your_key_here"}
    }
})

tools = await client.get_tools()
agent = create_agent("openai:gpt-4.1", tools)
response = await agent.ainvoke({"messages": "Summarize my last 5 emails"})

Step 1 β€” Install dependencies

pip install llama-index llama-index-tools-mcp

Step 2 β€” Connect and load tools

from llama_index.tools.mcp import MCPToolSpec
from llama_index.agent.openai import OpenAIAgent

mcp_spec = MCPToolSpec(
    server_url="https://mcp.toolforest.io/mcp",
    headers={"Authorization": "Bearer tfo_your_key_here"}
)

tools = mcp_spec.to_tool_list()
agent = OpenAIAgent.from_tools(tools)
response = agent.chat("What's on my calendar tomorrow?")

Step 1 β€” Install the SDK

pip install openai-agents

Step 2 β€” Connect and load tools

from agents import Agent, Runner
from agents.mcp import MCPServerStreamableHttp

server = MCPServerStreamableHttp(
    url="https://mcp.toolforest.io/mcp",
    headers={"Authorization": "Bearer tfo_your_key_here"}
)

agent = Agent(
    name="My Toolforest Agent",
    instructions="You are a helpful assistant with access to my tools.",
    mcp_servers=[server]
)

async with server:
    result = await Runner.run(agent, "Create a task for my meeting tomorrow")

Step 1 β€” Install the MCP SDK

pip install mcp

Step 2 β€” Connect and list tools

from mcp.client.streamable_http import streamablehttp_client
from mcp import ClientSession

async with streamablehttp_client(
    url="https://mcp.toolforest.io/mcp",
    headers={"Authorization": "Bearer tfo_your_key_here"}
) as (read, write, _):
    async with ClientSession(read, write) as session:
        await session.initialize()
        tools = await session.list_tools()
        print(tools)  # your enabled Toolforest tools

Any MCP-compatible client can connect to Toolforest using:

MCP Server URL  : https://mcp.toolforest.io/mcp
Transport       : Streamable HTTP  (SSE also supported)
Auth header     : Authorization: Bearer tfo_your_key_here

Refer to your framework's MCP documentation for how to pass custom headers to a remote HTTP MCP server.

Toolkits

Browse our growing collection of toolkits. Click any toolkit to view setup instructions and full capabilities.

Google Workspace

Google Docs

Create, edit, and manage documents with comprehensive section-based editing and rich text formatting.

Setup guide included

Google Sheets

Automate and manage spreadsheets with comprehensive data manipulation, formatting, and analysis.

Setup guide included

Google Slides

Create, edit, and manage professional presentations with full control over slides, content, and styling.

Setup guide included

Gmail

Read, search, and send emails with AI assistance. Manage drafts, organize with labels, and handle attachments.

Google Calendar

Manage your calendar with AI assistance. View, create, and organize events, check availability, and RSVP to invitations.

Setup guide included

Google Tasks

Create, organize, and manage to-do lists with due dates, subtasks, and progress tracking.

Health & Fitness

My Fitbit Data

Access health and fitness data with daily summaries, intraday tracking, trend analysis, and anomaly detection.

Setup guide included

Ōura Ring

Access sleep, activity, and readiness data with daily summaries, intraday tracking, trend analysis, and anomaly detection.

Setup guide included

Prediction Markets

Kalshi

Access regulated prediction market data with event search, orderbooks, trade feeds, and OHLC price history.

Setup guide included

Polymarket

Access prediction market data with market search, real-time orderbooks, and historical price analysis.

Setup guide included

Social

Twitter / X

Search tweets, explore trending topics, and analyze user profiles and engagement.

Setup guide included

Other

Apple Music

Get listening data, discover new music, and create custom playlists with AI assistance.

GitHub

Manage repositories, issues, pull requests, and projects with AI assistance.

FAQ

What is MCP?

MCP (Model Context Protocol) is an open protocol that standardizes how AI assistants connect to external tools and data sources. It was created by Anthropic and is supported by many LLM providers.

Do you store my data?

No. toolforest.io acts as a secure bridge between your LLM and your connected services. We don't permanently store your documents, emails, or other data. Data passes through our servers only for the duration of each operation and is immediately discarded.

Can I revoke access?

Yes! You can revoke access to any connected service at any time from your dashboard. You can also revoke access directly from the service provider (e.g., Google Account settings).

Which LLMs are supported?

toolforest.io can work with any LLM chatbot or agent that supports the Model Context Protocol.

Can I request new integrations?

Absolutely! Submit a feature request and let us know what you'd like to see.