Low Level MCP Server Streamable HTTP

You run a TypeScript-based MCP server that stores simple text notes with URIs and metadata, and exposes tools to create notes and prompts to summarize them. It provides a low‑level, streamable HTTP MCP endpoint plus a local runtime you can operate and test from your machine or Claude Desktop integration.

How to use

Access your notes through the note:// URIs, view titles, content, and metadata, and fetch the plain text content with a simple, consistent MIME type. Use the create_note tool to add new notes with a title and content, and invoke summarize_notes to generate a structured prompt that summarizes all stored notes. This setup lets you build quick note-taking workflows and drive LLM-based summaries directly from your MCP client.

How to install

Prerequisites
- Node.js installed on your machine (version 14+ recommended)
- npm (comes with Node.js)
- Optional: if you plan to debug with Claude Desktop, you may add the provided config snippet for your environment.

Install dependencies and build the server
- npm install
- npm run build

Run the server
- npm start

Configure Claude Desktop to connect to this MCP server by adding a JSON config snippet for mcpServers. For MacOS or Windows, use the paths shown to place the config file in the appropriate folder.

Example Claude Desktop MCP configuration block to copy into your config file:

{
  "mcpServers": {
    "streamable_http": {
      "type": "http",
      "url": "http://localhost:3001/mcp",
      "args": []
    }
  }
}

Additional notes

If you want to debug live MCP interactions, you can run the MCP Inspector tool to view a debugging URL in your browser. This helps verify the MCP connection and activity while you develop.

Troubleshooting and tips

Common issues include the server not starting on the expected port. Check that npm install completed successfully, the build finished without errors, and that no other process is occupying port 3001. If you encounter missing tool endpoints, ensure you started the correct server instance and that you are connecting via the note:// resources as described.

Available tools

create_note

Creates new text notes with mandatory title and content, storing the note in the server state.

summarize_notes

Generates a structured prompt to summarize all stored notes, embedding all note contents as resources.