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

# MCP integration

> Learn how to integrate Model Context Protocol (MCP) servers with FlowX.AI to enable AI agents to use external tools and services.

## Overview

The Model Context Protocol (MCP) integration connects FlowX.AI with external MCP servers and exposes their tools to AI agents. AI agents can interact with external systems, databases, APIs, and services through this protocol.

<Info>
  MCP is a protocol for AI agents to access external tools and data sources.
</Info>

## Key features

* Connect MCP servers as data sources in Integration Designer
* Enable or turn off specific tools exposed by MCP servers
* Use MCP tools in workflow processes through Custom Agent nodes
* Secure connections with multiple authentication methods
* Monitor tool usage and execution details in console logs

## How it works

```mermaid theme={"system"}
graph LR
    A[Integration Designer] -->|Connect| B[MCP Server]
    B -->|Expose Tools| C[Tool Registry]
    C -->|Select Tools| D[Custom Agent Node]
    D -->|Execute| E[AI Service]
    E -->|Use Tools| B
```

### High-level workflow

1. **Add MCP server**: Configure an MCP server connection as a data source in Integration Designer
2. **Authorize access**: Authenticate the connection using supported authentication methods
3. **Manage tools**: View and enable or turn off available tools from the MCP server
4. **Use in workflows**: Add Custom Agent nodes to workflows and select which MCP servers to use
5. **Execute and monitor**: Run workflows and track tool usage in console logs

## Supported protocols and features

<Info>
  The MCP integration is actively being developed. The features listed below represent the current capabilities. Additional protocol support and authentication methods will be added in future releases.
</Info>

### Transport protocols

The following transport protocols are currently supported:

<Accordion title="SSE (Server-Sent Events)">
  Server-Sent Events protocol for real-time, unidirectional communication from the MCP server to FlowX.AI.

  **Use Case**: Ideal for streaming responses and long-running operations where the server needs to push updates to the client.
</Accordion>

<Accordion title="Streamable HTTP">
  HTTP-based streaming protocol for receiving data in chunks over a persistent connection.

  **Use Case**: Suitable for large data transfers and scenarios where you need progressive response handling.
</Accordion>

### Proxy support

MCP connections support inline proxy configuration for environments that require routing through a proxy server. This is configured directly on each MCP data source connection.

| Setting                | Description                                             |
| ---------------------- | ------------------------------------------------------- |
| **Protocol**           | HTTP, HTTPS, SOCKS4, or SOCKS5                          |
| **Host and port**      | Proxy server address                                    |
| **Authentication**     | Optional basic authentication for the proxy             |
| **No-proxy patterns**  | Bypass list for addresses that should not use the proxy |
| **SSL verification**   | Toggle SSL/TLS certificate verification                 |
| **Connection timeout** | Configurable timeout for proxy connections              |

<Note>
  This is separate from the reusable [Proxy connections](../proxy-connections) resource available for RESTful System data sources. MCP data sources use their own inline proxy configuration.
</Note>

***

## Authentication methods

MCP integrations support the following authentication methods:

| Method                        | Description                                            | Use case                                     |
| ----------------------------- | ------------------------------------------------------ | -------------------------------------------- |
| **None**                      | No authentication                                      | Public MCP servers, development environments |
| **Basic Auth**                | Username and password                                  | Simple authentication requirements           |
| **Bearer Token**              | Static token-based authentication                      | API key or pre-generated token access        |
| **API Key**                   | Key-based authentication via header or query parameter | Third-party API access                       |
| **OAuth2 Client Credentials** | Service account authentication                         | Production service-to-service communication  |
| **OAuth2 User**               | User-context OAuth2 with redirect URI                  | Operations requiring user identity context   |

<Info>
  The authentication type is selected during MCP server setup and can be discovered automatically during connection testing using the **Supported Auth Discovery** feature.
</Info>

## Use cases

### External API integration

Connect to external APIs and services through MCP tools. AI agents can fetch data, trigger actions, and integrate with third-party systems.

### Database operations

Query databases, retrieve records, and perform data operations through MCP tools.

### Custom business logic

Expose custom business logic as MCP tools for AI agents to execute.

### Multi-system orchestration

Use multiple MCP servers in a single Custom Agent node to orchestrate operations across different systems.

## Next steps

<CardGroup cols={2}>
  <Card title="MCP Data Sources" href="./adding-mcp-data-source" icon="database">
    Add, configure, and manage MCP servers and their tools
  </Card>

  <Card title="Custom Agent Node" href="../custom-agent-node" icon="robot">
    Use MCP tools in workflows with Custom Agent nodes
  </Card>

  <Card title="MCP API Reference" href="../../../api/mcp-api-reference" icon="code">
    API documentation for MCP integration
  </Card>
</CardGroup>

## Related resources

<Card title="Integration Designer" href="../../integration-designer/designer-overview" icon="link">
  Integration Designer and data sources
</Card>

<Card title="Integrations Overview" href="../../integrations-overview" icon="link">
  Integration ecosystem overview
</Card>