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

# MCP

> Integrate the OpenRegister API into your AI assistant or application

The OpenRegister MCP Server enables AI assistants and agent frameworks to interact with the OpenRegister API through the [Model Context Protocol](https://modelcontextprotocol.io/introduction). It exposes company search and data retrieval as tools that any MCP-compatible client can use.

There are two ways to connect: the **remote MCP server** (no local setup needed) or the **local npm package** (for full control).

## Remote MCP Server

The easiest way to connect — no local installation required. The server handles authentication via OAuth when you connect.

| Transport       | Endpoint                          |
| --------------- | --------------------------------- |
| SSE (legacy)    | `https://mcp.openregister.de/sse` |
| Streamable HTTP | `https://mcp.openregister.de/mcp` |

### Use in AI Assistants

Connect directly in your browser — no terminal or config files needed.

<Tabs>
  <Tab title="Claude.ai">
    Claude.ai supports MCP servers natively through its Integrations settings.

    1. Open [Claude.ai](https://claude.ai) and go to **Settings → Integrations**
    2. Click **+ Add** or **+ Add Custom Integration**
    3. Enter the URL: `https://mcp.openregister.de/sse`
    4. Follow the authorization flow to connect your API key
  </Tab>

  <Tab title="ChatGPT">
    ChatGPT supports MCP servers via its Connectors feature (requires Developer Mode).

    1. Open [ChatGPT](https://chatgpt.com) and go to **Settings → Advanced**
    2. Enable **Developer Mode**
    3. Go to **Settings → Connectors** → **+ Add Custom Connector**
    4. Enter the URL: `https://mcp.openregister.de/mcp`
    5. Follow the authorization flow to connect your API key
  </Tab>
</Tabs>

### Use in Developer Tools

Edit a config file and restart your client. Requires [Node.js 18+](https://nodejs.org).

<Tabs>
  <Tab title="Claude Desktop">
    Edit your Claude Desktop configuration file:

    * **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
    * **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`

    ```json theme={"dark"}
    {
      "mcpServers": {
        "openregister": {
          "command": "npx",
          "args": ["-y", "mcp-remote@latest", "https://mcp.openregister.de/sse"]
        }
      }
    }
    ```

    Restart Claude Desktop. Look for the hammer icon (🔨) to confirm the connection.
  </Tab>

  <Tab title="Cursor">
    Edit `~/.cursor/mcp.json`:

    ```json theme={"dark"}
    {
      "mcpServers": {
        "openregister": {
          "command": "npx",
          "args": ["-y", "mcp-remote@latest", "https://mcp.openregister.de/sse"]
        }
      }
    }
    ```

    Restart Cursor to activate the connection.
  </Tab>
</Tabs>

### Build with It

Use the MCP server as a tool source in your own AI agents and pipelines.

<Tabs>
  <Tab title="OpenAI Agents SDK">
    ```python theme={"dark"}
    from agents import Agent, Runner
    from agents.mcp import MCPServerStreamableHttp

    async def main():
        async with MCPServerStreamableHttp(
            url="https://mcp.openregister.de/mcp",
            headers={"Authorization": "Bearer YOUR_API_KEY"},
        ) as mcp_server:
            agent = Agent(
                name="company-researcher",
                instructions="Use OpenRegister tools to look up German company data.",
                mcp_servers=[mcp_server],
            )
            result = await Runner.run(agent, "Find information about Trade Republic Bank GmbH")
            print(result.final_output)
    ```
  </Tab>

  <Tab title="Vercel AI SDK">
    ```typescript theme={"dark"}
    import { createOpenAI } from "@ai-sdk/openai";
    import { experimental_createMCPClient, generateText } from "ai";

    const mcpClient = await experimental_createMCPClient({
      transport: {
        type: "sse",
        url: "https://mcp.openregister.de/sse",
        headers: { Authorization: "Bearer YOUR_API_KEY" },
      },
    });

    const tools = await mcpClient.tools();

    const { text } = await generateText({
      model: createOpenAI({ apiKey: process.env.OPENAI_API_KEY })("gpt-4o"),
      tools,
      prompt: "Find information about Trade Republic Bank GmbH",
    });

    await mcpClient.close();
    ```
  </Tab>
</Tabs>

<Note>
  The remote MCP server requires Node.js 18 or higher for `mcp-remote`. If you encounter connection issues, try clearing the MCP auth cache: `rm -rf ~/.mcp-auth`
</Note>

***

## Local MCP Server (npm)

Run the MCP server locally with your API key for full control.

**Repository**: [openregister-typescript/packages/mcp-server](https://github.com/oregister/openregister-typescript/tree/main/packages/mcp-server)

```bash theme={"dark"}
export OPENREGISTER_API_KEY="your-api-key"
npx -y openregister-mcp@latest
```

For MCP clients that use configuration JSON (like Claude Desktop), add this configuration:

```json theme={"dark"}
{
  "mcpServers": {
    "openregister": {
      "command": "npx",
      "args": ["-y", "openregister-mcp", "--client=claude", "--tools=all"],
      "env": {
        "OPENREGISTER_API_KEY": "your-api-key"
      }
    }
  }
}
```

## Authentication

Set your OpenRegister API key as an environment variable:

```bash theme={"dark"}
export OPENREGISTER_API_KEY="your-api-key-here"
```

Get your API key from [OpenRegister Keys](https://openregister.de/keys).

For more detailed configuration options and advanced usage, see the [full repository documentation](https://github.com/oregister/openregister-typescript/tree/main/packages/mcp-server).