home / mcp / library mcp server

Library MCP Server

MCP server for operating on Markdown knowledge bases

Installation

Add the following to your MCP client configuration file.

Configuration

{
  "mcpServers": {
    "lethain-library-mcp": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/you/library-mcp",
        "run",
        "main.py",
        "/Users/you/content"
      ]
    }
  }
}

library-mcp is an MCP server that lets you interact with Markdown knowledge bases organized in folders containing .md files with YAML-style metadata. It enables you to retrieve recent content by tag or date range and discuss or summarize that content within an MCP-enabled client environment. This approach is well suited for working with large Markdown collections locally, across multiple knowledge bases, without uploading everything to a single context window.

How to use

You use an MCP client to connect to library-mcp and perform content queries or maintenance actions. Practical workflows include getting recent posts by a tag, then asking the system for recommendations or insights based on that subset, or summarizing posts within a specific date range. You can combine multiple queries to build a focused context before asking for analysis or decision support.

How to install

Prerequisites: you need a runtime capable of hosting an MCP server and access to a CLI toolchain that your chosen MCP client supports.

Clone the project to a local directory. Then install dependencies if necessary and start the server according to its runtime requirements.

# 1) Clone the project to a local path
git clone https://example.com/owner/library-mcp.git

# 2) Change to the project directory
cd library-mcp

# 3) Install dependencies if there is a package manifest (example shown for a typical Python setup)
pip install -r requirements.txt

# 4) Start the MCP server using the provided runtime command
uv --directory /path/to/library-mcp run main.py /path/to/content

Additional sections

Configuration example for running the MCP server locally is shown below. It runs the server using the uv runtime, pointing to a content directory that contains your Markdown files with metadata. Use the exact paths that fit your environment.

{
  "mcpServers": {
    "library": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/you/library-mcp",
        "run",
        "main.py",
        "/Users/you/content"
      ]
    }
  }
}

Available tools

get_by_tag

Retrieves content by tag from the knowledge base to bring relevant Markdown posts into your context window.

get_by_text

Searches content for specific text to locate relevant posts or passages.

get_by_slug_or_url

Finds posts by slug or URL for precise retrieval.

get_by_date_range

Gets posts published within a specified date range.

search_tags

Searches for tags matching a query to discover related topics.

list_all_tags

Lists all tags sorted by post count and recency.

rebuild

Rebuilds the content index to incorporate new or edited Markdown files.