Runtype
BlogSurfaces

Turn Your AI Into a Tool for Cursor, Claude Code, and Windsurf

Runtype's MCP surface lets you expose your flows and agents as tools that AI coding assistants can call directly. Your AI inside their AI.

Runtype TeamRuntype Team
mcpdeveloper-toolscursorclaude-codeintegration

Your developers already live inside AI coding assistants. Cursor, Claude Code, Windsurf. These tools are where they think, write, and ship.

What if your AI product could meet them there?

MCP: the protocol that connects AI tools

The Model Context Protocol (MCP) is a standard for AI tools to expose capabilities to other AI systems. When your agent speaks MCP, it becomes a tool that coding assistants can discover and call natively.

Runtype's MCP surface makes this happen with zero protocol implementation on your part.

Your AI inside their AI

Build your flow or agent in Runtype. Attach an MCP surface. You get an MCP-compatible endpoint that coding assistants can connect to.

From the user's perspective, they add your tool to their Cursor or Claude Code configuration. Then they can invoke it naturally, as part of their coding workflow.

MCP server configuration
{
"mcpServers": {
  "your-agent": {
    "url": "https://api.runtype.com/mcp/your-surface-id",
    "headers": {
      "Authorization": "Bearer YOUR_KEY"
    }
  }
}
}

Your agent appears as a tool. The coding assistant can call it when it's relevant. No context switching, no browser tabs, no copy-pasting.

Built-in MCP servers

Runtype also comes with built-in MCP integrations for common services:

  • Slack: Read channels, send messages, manage threads
  • Google Workspace: Access documents, emails, calendar
  • Linear: Create and manage issues, track projects
  • GitHub: Browse repos, manage PRs, read code

These are available as tools in your flows and agents. Your AI can read a Linear ticket, check the related GitHub PR, summarize the Slack discussion, and draft a response. All through MCP.

Custom MCP servers

Beyond the built-ins, Runtype supports connecting to any MCP server via URL. If a service publishes an MCP endpoint, you can plug it into your flows. Runtype handles auto-discovery of available tools and their parameter schemas.

Use cases

Internal tools as AI capabilities: Your team's knowledge base, deployment system, or analytics dashboard exposed as tools that any developer's coding assistant can use.

Product APIs for AI consumption: Your product's API, wrapped in natural language descriptions and parameter schemas, available to any MCP-compatible client.

AI-powered code generation: An agent that understands your codebase conventions, API patterns, and business logic, callable directly from the editor.

Auth modes

MCP surfaces support three authentication modes:

  • Keys only: Simple API key authentication
  • Required: Full authentication with user identity
  • None: Public access (for open tools or demos)

The Ship tab in Runtype generates installation instructions for Claude, Cursor, and Windsurf, with your specific authentication already configured.

Attach an MCP surface to any flow, copy the server URL into your mcp.json, and your agent shows up as a native tool in Cursor or Claude Code.