Knowledge MCP Server

A Model Context Protocol (MCP) server that provides centralized knowledge management for your projects. Store, search, and maintain project-specific knowledge that persists across sessions.

Features

Project-aware: Automatically identifies projects by git remote URL
Structured knowledge: Organize information in YAML-frontmatter markdown documents
Full-text search: Search across all project knowledge documents
Chapter-based organization: Structure documents with chapters for easy navigation
Main instructions: Replace CLAUDE.md with centralized project instructions
Secure: Path validation, input sanitization, and safe file operations

Installation

Via npx (recommended)

npx @spothlynx/knowledge-mcp

Via npm (global installation)

npm install -g @spothlynx/knowledge-mcp
knowledge-mcp

Requirements

Node.js 14+
Python 3.11+
Git (for project identification)

Configuration

Claude Desktop

Add to your Claude Desktop configuration (claude_desktop_config.json):

{
  "mcpServers": {
    "knowledge": {
      "command": "npx",
      "args": ["-y", "@spothlynx/knowledge-mcp"]
    }
  }
}

VSCode Claude Extension

Add to your VSCode settings:

{
  "claude.mcpServers": {
    "knowledge": {
      "command": "npx",
      "args": ["-y", "@spothlynx/knowledge-mcp"]
    }
  }
}

Other MCP Clients

# Direct execution
npx @spothlynx/knowledge-mcp

# Or if installed globally
knowledge-mcp

Usage

The Knowledge MCP Server provides the following tools:

Managing Project Instructions

Get project instructions: Retrieve the main project instructions (replaces CLAUDE.md)
Update project instructions: Create or update main project instructions

Knowledge Documents

Create knowledge file: Create structured documents with metadata and chapters
Search knowledge: Full-text search across all project documents
Update chapter: Modify specific chapters within documents
Delete knowledge file: Remove knowledge documents

Resources (Read-only)

knowledge://projects/{project_id}/main - Read main project instructions
knowledge://projects/{project_id}/files - List all knowledge files
knowledge://projects/{project_id}/chapters/{filename} - List chapters in a document

Storage Location

Knowledge is stored in your home directory:

~/.knowledge-mcp/
└── projects/
    └── {git-remote-url}/      # Project identified by git remote
        ├── main.md            # Main project instructions
        └── knowledge/         # Knowledge documents
            └── *.md           # Individual knowledge files

Document Format

Knowledge documents use YAML frontmatter with markdown content:

---
title: API Documentation
keywords: [api, rest, endpoints]
---

# Introduction
Overview of the API...

## Chapter: Authentication
How to authenticate with the API...

## Chapter: Endpoints
Available API endpoints...

Development

Setup Development Environment

# Clone the repository
git clone https://github.com/sven-borkert/knowledge-mcp.git
cd knowledge-mcp

# Create virtual environment
python3 -m venv venv
source venv/bin/activate

# Install in development mode with all dependencies
pip install -e ".[dev]"

# Run tests
pytest tests/ -v

# Run code quality checks
python check_code_quality.py --fix

Testing with MCP Inspector

# Test with MCP Inspector
npx @modelcontextprotocol/inspector python -m knowledge_mcp.server

# Or if installed
npx @modelcontextprotocol/inspector knowledge-mcp

Building and Distribution

# Build distribution packages
python -m build

# Upload to PyPI (when ready)
twine upload dist/*

Security

Path traversal protection
Input sanitization for filenames
Safe YAML parsing (no code execution)
Atomic file writes
Secure project identification

Troubleshooting

Python not found

# Ensure Python 3.11+ is installed
python3 --version

# On Windows
python --version

Permission errors

# Check directory permissions
ls -la ~/.knowledge-mcp/

# Create directory if needed
mkdir -p ~/.knowledge-mcp/projects

MCP connection issues

# Test with MCP Inspector
npx @modelcontextprotocol/inspector knowledge-mcp

# Check server logs
knowledge-mcp 2>&1 | tee server.log

Contributing

Fork the repository
Create a feature branch
Run tests and code quality checks
Submit a pull request

License

MIT License - see LICENSE file for details

Support

GitHub Issues: https://github.com/sven-borkert/knowledge-mcp/issues
MCP Documentation: https://modelcontextprotocol.io