> For the complete documentation index, see [llms.txt](https://docs.n8n.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.toolmcp.md).

# MCP Client Tool

Learn how to use the MCP Client Tool node in n8n. Follow technical documentation to integrate MCP Client Tool node into your workflows.

The MCP Client Tool node is a [Model Context Protocol (MCP)](https://modelcontextprotocol.io/introduction) client, allowing you to use the tools exposed by an external MCP server. You can connect the MCP Client Tool node to your models to call external tools with n8n agents.

{% hint style="info" %}
**Already in the registry?**

If the service you want is available in n8n's [MCP servers](/build/integrate-ai/mcp-servers.md) registry, you can connect it straight from the node panel without adding a credential. Use this node for MCP servers that aren't in the registry yet.
{% endhint %}

## Credentials

The MCP Client Tool node supports [Bearer](/integrations/builtin/credentials/httprequest.md#using-bearer-auth), generic [header](/integrations/builtin/credentials/httprequest.md#using-header-auth), multiple headers, and [OAuth2](/integrations/builtin/credentials/mcp.md#using-oauth2) authentication methods.

## Node parameters <a href="#node-parameters" id="node-parameters"></a>

Configure the node with the following parameters.

* **SSE Endpoint**: The SSE endpoint for the MCP server you want to connect to.
* **Authentication**: The authentication method for authentication to your MCP server. The MCP tool supports [bearer](/integrations/builtin/credentials/httprequest.md#using-bearer-auth), generic [header](/integrations/builtin/credentials/httprequest.md#using-header-auth), multiple headers, and [OAuth2](/integrations/builtin/credentials/mcp.md#using-oauth2) authentication. Select **None** to attempt to connect without authentication.
  * **Multiple Headers Auth**: Use this when your MCP server requires more than one header, for example an API key and a username. Add each header as a **Name** and **Value** pair in the credential. You can add as many headers as you need.
* **Tools to Include**: Choose which tools you want to expose to the AI Agent:
  * **All**: Expose all the tools given by the MCP server.
  * **Selected**: Activates a **Tools to Include** parameter where you can select the tools you want to expose to the AI Agent.
  * **All Except**: Activates a **Tools to Exclude** parameter where you can select the tools you want to avoid sharing with the AI Agent. The AI Agent will have access to all MCP server's tools that aren't selected.

## Templates and examples <a href="#templates-and-examples" id="templates-and-examples"></a>

[Browse MCP Client Tool node documentation integration templates](https://n8n.io/integrations/mcp-client-tool) or [search all templates](https://n8n.io/workflows/)

## Related resources <a href="#related-resources" id="related-resources"></a>

n8n also has an [MCP Server Trigger](/integrations/builtin/core-nodes/n8n-nodes-langchain.mcptrigger.md) node that allows you to expose n8n tools to external AI Agents.

Refer to the [MCP documentation](https://modelcontextprotocol.io/introduction) and [MCP specification](https://modelcontextprotocol.io/specification/) for more details about the protocol, servers, and clients.

Refer to [LangChain's documentation on tools](https://langchain-ai.github.io/langgraphjs/how-tos/tool-calling/) for more information about tools in LangChain.

View n8n's [Advanced AI](/build/integrate-ai.md) documentation.