Migrate from Heroku and get up to $10k in credits
Get startedWhy deploy the MCP server Typescript on Render?
MCP Server TypeScript is a template for building Model Context Protocol servers using the official TypeScript SDK with HTTP transport. It provides a starting point for developers to create MCP-compatible servers that expose custom tools to AI clients, with built-in authentication and deployment configuration for Render.
This template gives you a production-ready MCP server with bearer token authentication, health checks, and Streamable HTTP transport already configured—Render auto-generates your MCP_API_TOKEN on deploy so you're secured from the start. Instead of wiring up Express middleware, authentication logic, and deployment configs yourself, you get a working server in one click that's ready to connect to Cursor, Claude Desktop, or any MCP client. Fork it, add your tools, push—Render handles the rest.
Architecture
What you can build
After deploying, you'll have a live MCP server with token authentication that you can connect to Claude Desktop, Cursor, or Codex as a tool provider. It includes one example tool and an AGENTS.md file, so you can immediately start adding custom tools—either manually or by prompting an AI coding assistant to scaffold them for you.
Key features
- Streamable HTTP Transport: Uses the MCP TypeScript SDK with streamable HTTP transport, compatible with Cursor, Claude Desktop, and Codex clients.
- Bearer Token Auth: Auto-generates MCP_API_TOKEN on deploy with authentication middleware that can be disabled locally by unsetting the variable.
- Zod Schema Validation: Tool input schemas use Zod for type-safe parameter validation with defaults support in the registerTool API.
- One-Click Render Deploy: Includes render.yaml Blueprint for infrastructure-as-code deployment with health check endpoint at /health.
- AI Assistant Instructions: Ships with AGENTS.md so Cursor, Copilot, and Codex can scaffold new tools following project conventions automatically.
Use cases
- Solo dev deploys custom API tools for Claude Desktop in minutes
- Team lead scaffolds MCP tools using AI coding assistants with AGENTS.md
- Startup founder connects internal data sources to Cursor via authenticated endpoints
- Backend engineer prototypes LLM tool integrations locally before deploying to Render
What's included
Service | Type | Purpose |
|---|
Next steps
- Open the Render Dashboard, go to your service's Environment tab, and copy the MCP_API_TOKEN value — You should see a randomly generated token string that was auto-created during deployment
- Configure your MCP client (Cursor, Claude Desktop, or Codex) with your service URL and token using the examples in the README — When you save the config and restart the client, the server should appear in your available MCP servers list
- Test the hello tool by asking your MCP client to use it — You should receive a greeting response confirming the server is responding to tool calls correctly