# MCP Server Setup Workflow
Follow these steps to set up an MCP server for AI coding agents:
1. **Define tools** — List the capabilities your MCP server will expose (search, fetch, execute).
2. **Scaffold the server** — Use the MCP SDK for TypeScript or Python with typed tool schemas.
3. **Implement handlers** — Write safe, validated handlers for each tool with clear error messages.
4. **Configure transport** — Choose stdio for local editors or SSE/HTTP for remote access.
5. **Add to your editor** — Register the server in Cursor MCP settings, Claude Code, or Windsurf MCP config.
6. **Test end-to-end** — Verify each tool works from the agent chat with realistic inputs.
7. **Document usage** — Add a README with setup commands, env vars, and example prompts.
Follow as a workflow guide for MCP server setup and integration.