All pages
Powered by GitBook
1 of 1

Loading...

AI Agents for MCP

Learn about creating an MCP server with MindStudio.

What is an MCP Server?

An MCP (Model Context Protocol) server lets you expose one or more of your MindStudio agents as tools that external applications or LLMs can invoke. In practice, an MCP server:

  • Enables agents to work with external tools and data sources without having to rebuild or redeploy them each time.

  • Serves as a centralized endpoint that routes requests to any published agent you’ve added.


How to Build an MCP Server with MindStudio

  1. Configure Your Agent as a

    • Open the editor of agent you want to expose and select the Start block.

    • Change the trigger type to Packaged Workflow.

  1. Save and Publish

    • Click Save to persist your changes.

    • Click Publish so the workflow is available in the workspace.

Connection Details Example Snippet


Integrating Your MCP Server with External Tools

  • Copy the connection details snippet into the desired external application. This will vary depending on the service you are using.

Fill in Workflow Metadata
Field
Example Value

Workflow ID/Name

Get AI News Scrape

Description

Get a bullet point summary of today’s AI news.

Inputs

(None—this workflow runs standalone)

Outputs

One output: the summary string

Copy Your Agent’s ID
  • After publishing, the agent’s general settings will show its Agent ID.

  • Click the copy icon next to that ID

  • Create the MCP Server

    • Navigate back to your main MindStudio workspace.

    • Select MCP Server from the Workspace settings dropdown.

    • Paste in your copied Agent ID in the space provided.

    • Tap Add to register this agent on the server.

    • Copy the connection details snippet to be used with your external integration.

  • Packaged Workflow
    Claude
    Open AI
    Cursor
    {
      "mcpServers": {
        "mindstudio": {
          "command": "npx",
          "args": [
            "mcp-remote",
            "https://v1.mindstudio-api.com/developer/mcp/v1/YOUR-AGENT-ID",
            "--header",
            "Authorization: ${AUTH_TOKEN}",
            "--transport http-only"
          ],
          "env": {
            "AUTH_TOKEN": "Bearer YOUR_API_KEY"
          }
        }
      }
    }