> ## Documentation Index
> Fetch the complete documentation index at: https://docs.logic.inc/llms.txt
> Use this file to discover all available pages before exploring further.

# MCP Quick Start

> Create your first Logic MCP server in minutes

## Creating Your First MCP Server

This guide will walk you through creating an MCP server and connecting it to an AI platform.

### Prerequisites

Before you begin, make sure you have:

* Organization administrator access (for creating MCP servers)
* An MCP-compatible AI platform (such as Claude, ChatGPT, or Gemini)

## Step 1: Navigate to MCP Server Settings

1. Click the **gear icon** in the bottom left corner of the Logic interface to open Organization Settings
2. Navigate to the "MCP Servers" tab

<img src="https://mintcdn.com/logic/YO7O6hkpncKkRwwK/screenshots/MCP_servers.png?fit=max&auto=format&n=YO7O6hkpncKkRwwK&q=85&s=605cdb15c3559d2e0fe6e8b5f7a6004d" alt="MCP servers settings page" width="3840" height="2160" data-path="screenshots/MCP_servers.png" />

## Step 2: Create a New MCP Server

Click the **Create MCP Server** button to begin setup.

### Configure Your Server

You'll need to provide:

**Server Name** (required)

* A descriptive name for this MCP server (1-100 characters)
* Example: "Engineering Team Agents" or "Production Workflows"

**Description** (optional)

* Additional context about what this server provides (up to 500 characters)
* This helps team members understand the server's purpose

**Agent Scope** (required)

* Choose how this server accesses your agents:
  * **Full Organization**: All current and future agents in your organization
  * **Limited**: Only specific agents you select

<Info>
  Not sure which scope to choose? See the [Agent Scope](/mcp/agent-scope)
  guide for detailed recommendations.
</Info>

### Select Agents (Limited Scope Only)

If you chose Limited scope, select which agents this server can access. You can:

* Search for agents by name
* Select multiple agents
* Add or remove agents later

<img src="https://mintcdn.com/logic/YO7O6hkpncKkRwwK/screenshots/MCP_servers_select_agents.png?fit=max&auto=format&n=YO7O6hkpncKkRwwK&q=85&s=52d0e51e0be8a0869474b681b83ed338" alt="MCP servers select which agents are accessible" width="3840" height="2160" data-path="screenshots/MCP_servers_select_agents.png" />

## Step 3: Copy Your Server URL

After creating the server, Logic generates a unique connection URL that AI platforms use to connect to your MCP server.

Copy the URL - you'll need it to configure your AI platform in the next step.

## Step 4: Connect to an AI Platform

Now that you have your MCP server URL, connect it to an AI platform:

<CardGroup cols={2}>
  <Card title="Claude Desktop" icon="robot" href="/mcp/integrations/claude-desktop">
    Configure Claude desktop app
  </Card>

  <Card title="Claude Code" icon="terminal" href="/mcp/integrations/claude-code">
    Set up Claude Code CLI
  </Card>

  <Card title="Codex" icon="code" href="/mcp/integrations/codex">
    Set up Codex CLI
  </Card>

  <Card title="ChatGPT" icon="comments" href="/mcp/integrations/chatgpt">
    Set up ChatGPT with developer mode
  </Card>
</CardGroup>

## Step 5: Test Your Connection

Once connected, test that your AI platform can access your Logic agents:

1. Open your AI platform
2. Look for your Logic agents in the available tools/functions
3. Try executing a simple Logic agent
4. Verify you receive the expected response

## Next Steps

Now that your MCP server is running:

* **Learn about [Agent Scope](/mcp/agent-scope)** - Optimize which agents are accessible
* **Explore [Async Tasks](/mcp/async-tasks)** - Handle long-running workflows efficiently

<Note>
  Need help? Check the platform-specific integration guides for detailed setup
  instructions and troubleshooting.
</Note>