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

# n8n

> Connect KonbiniAPI's hosted MCP server to n8n workflows and AI agents.

n8n supports MCP as both a client and a server. For KonbiniAPI, the useful part
is n8n's MCP client support, which lets your workflows and AI agents call
KonbiniAPI tools from the hosted remote MCP.

## Use KonbiniAPI in regular workflows

Use the **MCP Client** node if you want MCP tools to behave like normal steps in
an n8n workflow.

<Steps>
  <Step title="Add an MCP Client node">
    In your workflow, add the **MCP Client** node.
  </Step>

  <Step title="Set the remote MCP URL">
    Configure:

    * **MCP Endpoint URL**: `https://mcp.konbiniapi.com`
    * **Server Transport**: choose the HTTP transport your n8n version exposes for MCP client connections
  </Step>

  <Step title="Authenticate">
    In the node credentials, choose one of these:

    * **OAuth2** if you want n8n to complete the browser-based OAuth flow
    * **Bearer** if you want to use a KonbiniAPI API key directly
  </Step>

  <Step title="Pick the tool and map inputs">
    Select the KonbiniAPI tool you want to call, then map the inputs from your
    workflow data.
  </Step>
</Steps>

## Use KonbiniAPI with n8n AI agents

Use the **MCP Client Tool** node if you want KonbiniAPI to appear as tools for
an n8n AI Agent.

<Steps>
  <Step title="Add an MCP Client Tool node">
    Add **MCP Client Tool** and connect it to your AI Agent.
  </Step>

  <Step title="Set the endpoint">
    Use `https://mcp.konbiniapi.com` as the MCP server URL.
  </Step>

  <Step title="Choose auth">
    Use **OAuth2** or **Bearer** credentials, depending on whether you want an
    OAuth login flow or a direct KonbiniAPI API key.
  </Step>

  <Step title="Expose the tools">
    Include all tools, or select only the KonbiniAPI tools you want the agent to
    use.
  </Step>
</Steps>

## Recommended KonbiniAPI pattern

* Pick the typed tool that matches the task, such as `instagram_get_user` or `tiktok_get_user_videos`
* Use `projection_preset`, `data_fields`, and `item_fields` to keep results small
* Parse the returned JSON string in the next step if your workflow needs fields individually

Official sources:

* [n8n MCP Client node docs](https://docs.n8n.io/integrations/builtin/core-nodes/n8n-nodes-langchain.mcpClient/)
* [n8n MCP Client Tool node docs](https://docs.n8n.io/integrations/builtin/cluster-nodes/sub-nodes/n8n-nodes-langchain.toolmcp/)