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

# MCP Server

> Connect AI assistants like Claude, Cursor, and OpenCode to your Inbound account

The Inbound MCP (Model Context Protocol) server allows AI assistants to interact with your Inbound account to manage domains, endpoints, and emails.

## Quick Start

The fastest way to get started is using Inbound's hosted MCP server at `mcp.inbound.new/mcp`.

### Cursor

Add to your Cursor MCP config (`.cursor/mcp.json` in your project or global config):

```json theme={null}
{
  "mcpServers": {
    "inbound": {
      "type": "http",
      "url": "https://mcp.inbound.new/mcp",
      "headers": {
        "x-inbound-api-key": "your-api-key"
      }
    }
  }
}
```

### OpenCode

Add to your `opencode.json` config:

```json theme={null}
{
  "mcp": {
    "inbound": {
      "type": "remote",
      "url": "https://mcp.inbound.new/mcp",
      "headers": {
        "x-inbound-api-key": "your-api-key"
      }
    }
  }
}
```

### Claude Desktop

Add to your Claude Desktop config (`~/Library/Application Support/Claude/claude_desktop_config.json` on macOS):

```json theme={null}
{
  "mcpServers": {
    "inbound": {
      "command": "npx",
      "args": ["-y", "@inbound/mcp"],
      "env": {
        "INBOUND_API_KEY": "your-api-key"
      }
    }
  }
}
```

<Note>
  Claude Desktop uses STDIO transport and runs the MCP server locally via npx.
</Note>

## Authentication

The MCP server authenticates using your Inbound API key. Get your API key from the [dashboard](https://inbound.new/settings).

### Domain Restriction (Optional)

To restrict all operations to a single domain (and its subdomains), add the `x-inbound-domain` header:

```json theme={null}
{
  "mcpServers": {
    "inbound": {
      "type": "http",
      "url": "https://mcp.inbound.new/mcp",
      "headers": {
        "x-inbound-api-key": "your-api-key",
        "x-inbound-domain": "example.com"
      }
    }
  }
}
```

This will filter results to show only `example.com` and subdomains like `mail.example.com`, `support.example.com`, etc.

## Available Tools

### Domains

| Tool           | Description                                                    |
| -------------- | -------------------------------------------------------------- |
| `list_domains` | List all domains in your account (respects domain restriction) |

### Endpoints

| Tool              | Description                                              |
| ----------------- | -------------------------------------------------------- |
| `list_endpoints`  | List webhook and email forwarding endpoints              |
| `create_endpoint` | Create a webhook, email forward, or email group endpoint |

### Emails

| Tool          | Description                                                            |
| ------------- | ---------------------------------------------------------------------- |
| `list_emails` | List sent, received, and scheduled emails with filtering               |
| `get_email`   | Get detailed information about a specific email including full content |
| `send_email`  | Send or schedule an email                                              |

### Threads

| Tool           | Description                            |
| -------------- | -------------------------------------- |
| `list_threads` | List email conversations with previews |
| `get_thread`   | Get all messages in a thread           |

## Example Usage

Once configured, you can ask your AI assistant to:

* "List all my domains"
* "Show me the last 10 received emails"
* "Send an email to [support@example.com](mailto:support@example.com) with subject 'Hello'"
* "Create a webhook endpoint for [notifications@mydomain.com](mailto:notifications@mydomain.com)"
* "Show me the conversation thread with [john@example.com](mailto:john@example.com)"
* "Get the full content of email ID abc123"

## Available Prompts

The MCP server includes a `getting-started` prompt that teaches the AI assistant how to use Inbound effectively. Most AI assistants will automatically discover and use this prompt.

## Self-Hosting

You can also run the MCP server locally:

### Installation

```bash theme={null}
npm install @inbound/mcp
# or
pnpm add @inbound/mcp
# or
bun add @inbound/mcp
```

### Running

```bash theme={null}
# HTTP server (for Cursor, OpenCode, etc.)
INBOUND_API_KEY=your-api-key npx @inbound/mcp start:http

# STDIO server (for Claude Desktop)
npx @inbound/mcp
```

The HTTP server runs on port 3002 by default. Configure your client to use `http://localhost:3002/mcp`.

## Troubleshooting

<AccordionGroup>
  <Accordion title="Tools not appearing">
    Make sure your API key is valid and has the necessary permissions. Check the MCP server logs for authentication errors.
  </Accordion>

  <Accordion title="Domain filtering not working">
    Ensure the `x-inbound-domain` header value matches exactly with your domain name in Inbound (case-sensitive).
  </Accordion>

  <Accordion title="Connection refused">
    If self-hosting, verify the server is running and the port is accessible. For the hosted version, check your network connectivity.
  </Accordion>
</AccordionGroup>

## Security

<Warning>
  Never commit your API key to version control. Use environment variables or a secrets manager.
</Warning>

Best practices:

* Use environment variables for API keys: `"x-inbound-api-key": "$INBOUND_API_KEY"`
* Restrict operations to specific domains using `x-inbound-domain`
* Regularly rotate your API keys
* Use separate API keys for different environments (dev, staging, production)