knowledge-base-mcp: Instructions file for Claude Code

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Project Overview

This is a Knowledge Base MCP (Model Context Protocol) Server - a mem0-like memory system for GitHub Copilot that provides persistent knowledge storage and retrieval using local ChromaDB. The server enables GitHub Copilot to save and retrieve contextual information about development environments.

Key Commands

Testing and Development

# Install dependencies
pip install -r requirements.txt

# Test server functionality
python test_server.py

# Test initial knowledge loading
python test_initial_knowledge.py

# Run usage examples
python examples.py

# Start the MCP server
python kb_server.py

No Build/Lint Commands

This is a pure Python project with no formal build, lint, or test runner configuration. Tests are run directly as Python scripts.

Architecture

Core Components

• kb_server.py: Main FastMCP server with ChromaDB integration
• ChromaDB: Local vector database for semantic search using all-MiniLM-L6-v2 embeddings
• FastMCP: MCP server framework for tool definitions

Key Tools

• kb_save: Save memories with automatic metadata extraction
• kb_search: Semantic search with vector similarity
• kb_list: List all saved memories with filtering
• kb_delete: Remove memories by ID

Memory Types

• Environment: URLs, configurations, dashboard locations
• Code Snippet: Code examples and patterns
• Operational: Troubleshooting steps and fixes
• Architectural: Design decisions and system patterns

Data Flow

1. Content saved via kb_save → metadata extraction → ChromaDB storage
2. Search queries → vector embedding → similarity search → ranked results
3. All data stored locally in kb_data/ directory

Configuration

Environment Variables

• KB_DATA_DIR: ChromaDB storage directory (default: ./kb_data)
• KB_INITIAL_FILE: Optional initial knowledge file to bootstrap database