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

# External MCP Servers

> Use BoostGPT MCP servers with other AI agents

## Overview

Your BoostGPT MCP servers can be used with any AI agent that supports the MCP protocol, not just BoostGPT AI teammates.

## Supported AI Agents

> [card] [BoostGPT AI Teammates](/docs/integrations/external/boostgpt-agents.md)
>
> Use with your BoostGPT AI teammates

> [card] [Claude Desktop](/docs/integrations/external/claude-desktop.md)
>
> Connect to Claude Desktop app

> [card] [ChatGPT](/docs/integrations/external/chatgpt.md)
>
> Use with ChatGPT (via plugins)

## How It Works

When you create an MCP server in BoostGPT, you get a custom URL:

```text theme={null}
https://mcp.your-server-id.boostgpt.co
```

This URL can be used with:

* **BoostGPT AI teammates** - Native integration
* **Claude Desktop** - MCP protocol support
* **ChatGPT** - Via custom plugin/action
* **Other MCP-compatible AI teammates** - Any AI teammate supporting MCP

## Getting Your MCP URL

**1. Create MCP Server**

**Dashboard -> MCP Servers -> Create Server**

**2. Get URL**

After creation, copy your MCP server URL:

```text theme={null}
https://mcp.abc123xyz.boostgpt.co
```

**3. Use Anywhere**

Connect this URL to any MCP-compatible AI agent

## Use Cases

### Internal Tools Everywhere

Create once, use everywhere:

```text theme={null}
1. Create MCP server from your internal API
2. Get MCP URL: https://mcp.internal-crm.boostgpt.co
3. Connect to:
   - BoostGPT teammates for customer support
   - Claude Desktop for personal use
   - Team's AI tools
```

### Team Collaboration

Share MCP servers across your team:

```text theme={null}
1. Create MCP server for company database
2. Team members connect to their preferred AI agents
3. Everyone has access to same tools
```

### Multi-Agent Workflows

Use different AI teammates for different tasks:

```text theme={null}
- BoostGPT AI teammate: Customer-facing chat
- Claude Desktop: Internal analysis
- Both use same MCP servers for data access
```

## Security

### Authentication

All MCP server requests require authentication:

* **API Keys** - Configured in MCP server settings
* **Rate Limiting** - Prevent abuse
* **Access Logs** - Monitor usage

### Permissions

Control who can use your MCP servers:

* **Dashboard -> MCP Servers -> Your Server -> Access**
* Add team members or make public
* Revoke access anytime

## Best Practices

1. **Name servers clearly** - Easy to identify purpose
2. **Document tools** - Add descriptions for each endpoint
3. **Monitor usage** - Check logs regularly
4. **Rotate credentials** - Update API keys periodically
5. **Version your APIs** - Support multiple MCP server versions

## Example: Internal CRM

Create an MCP server for your CRM and use it everywhere:

```text theme={null}
1. Dashboard -> MCP Servers -> Create Server
2. Import your CRM OpenAPI spec
3. Get URL: https://mcp.company-crm.boostgpt.co

Use with:
- BoostGPT support teammate
- Claude Desktop for sales team
- Custom internal tools
```

All AI teammates can now:

* Query customer data
* Create contacts
* Update deals
* Generate reports

## Next Steps

> [card] [BoostGPT AI Teammates](/docs/integrations/external/boostgpt-agents.md)
>
> Connect to BoostGPT AI teammates

> [card] [Claude Desktop](/docs/integrations/external/claude-desktop.md)
>
> Use with Claude Desktop

> [card] [ChatGPT](/docs/integrations/external/chatgpt.md)
>
> Connect to ChatGPT


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.