mcp-smart-notes

The mcp-smart-notes server provides a local, MCP-compliant note-taking system with automatic tagging powered by a local LLM via Ollama. It exposes a set of MCP tools to create, search, and manage notes with intelligent categorization. You can rely on the JSON-RPC 2.0 transport for communication with clients and leverage the built-in robust error handling to gracefully recover from common issues like model or Ollama unavailability. Typical workflows include creating a note with a title and content, after which the system automatically assigns tags from predefined categories such as Coding, Education, Greeting, and Finance.

Available tools include create_note (to add new notes with automatic tagging), search_notes (full-text search across titles, content, and tags), search_by_tag (find notes by a specific tag), list_notes (overview of notes with IDs and tagging status), update_note (modify title, content, or tags), and delete_note (permanently remove a note by ID). These tools enable a compact CLI or integration layer to drive a local knowledge base with privacy-preserving, on-device inference.

Prerequisites:

• Python 3.8 or newer
• Ollama installed and running locally
• A compatible local LLM model available to Ollama (recommended: qwen2.5:7b)

Installation steps:

1. Clone the repository git clone https://github.com/yourusername/mcp-smart-notes.git cd mcp-smart-notes

2. Install Python dependencies python -m pip install --upgrade pip pip install -r requirements.txt

3. Start Ollama in a separate terminal ollama serve

4. Pull a compatible model (in another terminal) ollama pull qwen2.5:7b

5. Run the MCP server python -m note_server

If you prefer a different invocation, adjust the command to point to your server module or entry point as needed.

Environment and configuration tips:

• Ensure Ollama is running and the chosen model is available locally to prevent failures in auto-tagging.
• The server uses JSON-RPC 2.0; clients should send properly structured requests and handle standard MCP error codes.
• If the model or Ollama becomes temporarily unavailable, the system will attempt fallbacks and report actionable errors rather than crashing.
• You can customize tag categories by adjusting the tagging bridge or rules within the codebase, provided you maintain MCP compatibility.
• For production deployments, consider running Ollama with sufficient resources and enable log levels that aid debugging without leaking sensitive note content.