todo

This MCP server provides a TODO management API with five operations: addTodo, getTodos, completeTodo, markIncomplete, and deleteTodo. It uses a Cloudflare Workers backend with a D1 database to persist TODO items. After you deploy, you can interact with the server through the MCP protocol exposed by the worker and test via the MCP Inspector during development. The server stores each TODO with fields like id, title, completed, created_at, and updated_at, and supports basic CRUD-style operations tailored for a simple task list.

During development and testing, you can start the local development server and then connect the MCP Inspector to verify behavior. The server is designed to be used in environments that support the MCP protocol, and the local workflow relies on Cloudflare tooling (wrangler) and D1 for persistence. After deployment, you can configure clients (e.g., Claude Desktop) to connect to your worker URL via the mcp-remote integration, enabling you to issue the standard TODO-related commands against your remote MCP server.

Prerequisites:

• Node.js (recommended LTS) and npm
• Cloudflare Wrangler CLI installed
• A Cloudflare account and access to create D1 databases

Step-by-step installation:

1. Create the MCP project from the template npm create cloudflare@latest -- my-mcp-server --template=cloudflare/ai/demos/remote-mcp-authless cd my-mcp-server

2. Create a D1 database for TODO tasks wrangler d1 create todo-tasks-db

3. Bind the D1 database in wrangler.jsonc Add the following binding to wrangler.jsonc (replace with your actual database_id): { "$schema": "node_modules/wrangler/config-schema.json", "name": "todo-mcp-server", "main": "src/index.ts", "compatibility_date": "2025-03-10", "compatibility_flags": ["nodejs_compat"], "d1_databases": [ { "binding": "DB", "database_name": "todo-tasks-db", "database_id": "your-database-id-here" } ] }

4. Create the database schema for TODOs Create a file called schema.sql with: CREATE TABLE todos ( id INTEGER PRIMARY KEY AUTOINCREMENT, title TEXT NOT NULL, completed BOOLEAN DEFAULT FALSE, created_at DATETIME DEFAULT CURRENT_TIMESTAMP, updated_at DATETIME DEFAULT CURRENT_TIMESTAMP );

5. Apply the schema locally or remotely wrangler d1 execute todo-tasks-db --file=./schema.sql --local wrangler d1 execute todo-tasks-db --file=./schema.sql --remote

6. Generate TypeScript types (if applicable) npm run cf-typegen

7. Run the development server locally npm run dev

8. Optional: Run the MCP Inspector to test locally npx @modelcontextprotocol/inspector@latest Open the port shown and navigate to http://localhost:8787/mcp to test

9. Deploy to Cloudflare npm run deploy

Notes and tips:

• The server uses a D1 database bound as DB in wrangler.jsonc; ensure the database_id is correct.
• If you modify schema, re-apply migrations with wrangler d1 execute todo-tasks-db --file=./schema.sql --remote (or --local for local testing).
• The MCP tools exposed by this server include addTodo, getTodos, completeTodo, markIncomplete, and deleteTodo. Ensure your MCP client is configured to call these operations according to the MCP protocol.
• For client configuration (e.g., Claude Desktop), point the mcpServers.todo to the deployed worker URL with the mcp-remote wrapper, as shown in the README: { "mcpServers": { "todo": { "command": "npx", "args": [ "mcp-remote@latest", "https://your-worker-domain.workers.dev/mcp" ] } } }
• If you see database binding issues, confirm that wrangler.jsonc includes the correct d1_databases[] binding name (DB) and that the database_id matches the created D1 instance.
• The project template demonstrates a remote MCP-authless setup; authentication and access controls can be layered on top if needed for production.
• This server is intended to run on a Cloudflare Workers environment; running locally relies on Wrangler and a compatible development workflow.