Notemd MCP

Notemd MCP is an MCP server that provides an AI-powered text processing and knowledge management backend for Markdown workspaces and Obsidian vaults. It connects AI models to Markdown documents, allowing knowledge workers, researchers, and Obsidian users to automate document structuring and cross-referencing. The server enables automated extraction of key concepts, dynamic creation of wiki-links, and generation of full structured documentation from simple titles. It also provides capabilities for web research through DuckDuckGo or Tavily, content summarization into Mermaid mindmaps, text translation, and automatic link maintenance when files are renamed or deleted. Built using FastAPI with a Node.js MCP bridge, Notemd MCP works with any OpenAI-compatible API, local models running through Ollama or LMStudio, and cloud providers like Anthropic and DeepSeek. By running computationally heavy knowledge base maintenance operations in the background, it ensures notes remain interconnected, syntax errors in Mermaid or LaTeX blocks are corrected, and duplicate concepts are filtered consistently.

Category: AI & LLM Tooling

Tags: knowledge-management, obsidian, plugin, text-processing

Visit Notemd MCP

How to install and configure Notemd MCP

To install and run the Notemd MCP server, choose one of the following methods: 1. Run directly using npx: bash npx notemd-mcp-server 2. Alternatively, clone the repository and install dependencies with Python: bash git clone https://github.com/Jacobinwwey/notemd-mcp.git cd notemd-mcp pip install -r requirements.txt uvicorn main:app --reload 3. To configure the server within an MCP client configuration file (such as Claude Desktop's settings.json), add the following entry to your mcpServers object: json { "mcpServers": { "notemd-mcp": { "description": "Notemd MCP Server - AI-powered text processing and knowledge management for your Markdown files.", "command": "npx", "args": [ "-y", "notemd-mcp-server" ], "env": { "OPENAI_API_KEY": "your_openai_api_key_here", "DEEPSEEK_API_KEY": "your_deepseek_api_key_here" } } } }

What you can do with Notemd MCP

• Enrich Markdown files automatically by parsing concepts and generating bidirectional wiki-links across your personal knowledge base. - Produce comprehensive reference documentation from a single topic title using web search queries and LLM generation. - Translate notes across multiple languages while preserving existing markdown formatting and core structural elements. - Generate Mermaid mindmaps and structured diagrams from raw note content for visual knowledge overviews. - Maintain vault link integrity by automatically updating or clearing backlinks when files are renamed or removed. - Batch-repair broken Mermaid.js diagram syntax and LaTeX mathematical equations across Markdown files within a directory.

Key facts

• https://github.com/Jacobinwwey/notemd-mcp
• AI & LLM Tooling, Notes & Knowledge Management
• knowledge-management, obsidian, plugin, text-processing

Part of MCP Servers

What is Notemd MCP?

Notemd MCP is a backend server that implements the Model Context Protocol to expose AI-driven text processing, note-linking, and knowledge management functionalities originally built for the Notemd Obsidian plugin.

How do I install Notemd MCP?

You can run it instantly using the command npx notemd-mcp-server. Alternatively, clone the GitHub repository, install requirements via pip or uv, and run uvicorn main:app --reload.

Which LLMs are supported by Notemd MCP?

The server supports any OpenAI-compatible API. This includes local model runners such as LMStudio and Ollama, as well as cloud-hosted providers including OpenAI, DeepSeek, Anthropic, and Google.

Which MCP clients work with Notemd MCP?

Any MCP-compatible client that supports external stdio processes, such as Claude Desktop, Cursor, or customized agent runners, can connect to Notemd MCP by adding it to their configuration file.

What search engines does Notemd MCP use for research?

The server includes integrated web research capabilities that can query either DuckDuckGo directly or use the Tavily API, configured via its central configuration settings.