Knowledge Base MCP Server

# Knowledge Base MCP Server - Setup Guide ## Quick Start 1. **Install dependencies**: ```bash pip install -r requirements.txt ``` 2. **Test the server**: ```bash python test_server.py ``` 3. **Run examples**: ```bash python examples.py ``` 4. **Start the server**: ```bash python kb_server.py ``` ## GitHub Copilot Integration ### For VS Code (Recommended) Add to your MCP configuration or VS Code settings: ```json { "mcpServers": { "knowledge-base": { "command": "python", "args": ["/absolute/path/to/kb-mcp/kb_server.py"], "env": { "KB_DATA_DIR": "/absolute/path/to/kb-mcp/kb_data", "KB_INITIAL_FILE": "/absolute/path/to/kb-mcp/initial_knowledge.txt" } } } } ``` ### For Claude Desktop (Testing) Copy `claude_desktop_config.json` to your Claude Desktop config directory and update the paths: **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json` **Windows**: `%APPDATA%\Claude\claude_desktop_config.json` ## Usage with GitHub Copilot ### Saving Memories In GitHub Copilot, you can save memories using the `kb_save` tool: ``` #kb_save we use splunk at https://company.splunkcloud.com for logging #kb_save when dynatrace fails, use DT_DISABLE flag and restart #kb_save here's our test pattern: ```python def test_api(): assert response.status_code == 200 ``` ``` ### Automatic Retrieval GitHub Copilot will automatically search your knowledge base when you ask questions: - "How do I check application logs?" → Returns Splunk dashboard info - "How to fix dynatrace issues?" → Returns troubleshooting steps - "Show me testing patterns" → Returns saved code snippets ### Manual Search You can also explicitly search: ``` #kb_search logging dashboard #kb_search graphql testing #kb_search troubleshooting steps ``` ## Features ✅ **Local Storage**: All data stored locally, no cloud dependencies ✅ **Semantic Search**: Vector-based similarity search using ChromaDB ✅ **Smart Metadata**: Auto-extracts technologies, URLs, and categorizes content ✅ **Memory Types**: Environment, Code Snippets, Operational, Architectural ✅ **Fast Performance**: Sub-500ms search with local embeddings ✅ **Corporate Compliant**: No external API calls required ## File Structure ``` kb-mcp/ ├── kb_server.py # Main MCP server ├── test_server.py # Functionality tests ├── test_initial_knowledge.py # Initial knowledge loading tests ├── examples.py # Usage examples ├── requirements.txt # Dependencies ├── initial_knowledge.txt # Example initial knowledge file ├── claude_desktop_config.json # Claude config template ├── README.md # Documentation ├── SETUP.md # This file ├── PRD-Knowledge-Base-MCP.md # Product requirements └── kb_data/ # ChromaDB storage (auto-created) ``` ## Troubleshooting ### Server Won't Start - Check Python version: `python --version` (requires 3.9+) - Install dependencies: `pip install -r requirements.txt` - Check permissions on kb_data directory ### Poor Search Results - Use specific keywords in searches - Save more descriptive content - Include technology names and URLs ### GitHub Copilot Not Finding Tools - Verify MCP configuration paths are absolute - Restart VS Code after configuration changes - Check that server starts without errors ## Advanced Configuration ### Environment Variables - `KB_DATA_DIR`: Custom data directory (default: `./kb_data`) - `KB_INITIAL_FILE`: Path to initial knowledge file (optional) ### Bootstrap with Initial Knowledge Create an `initial_knowledge.txt` file with your existing knowledge: ``` we use splunk at https://company.splunkcloud.com for logging our grafana dashboard is at https://monitoring.company.com when dynatrace fails, use DT_DISABLE flag and restart ``` Entries should be separated by double newlines. The server will automatically load these on startup. ### Custom Embedding Models The server uses ChromaDB's default embeddings. To use different models, modify the embedding function in `init_database()`. ## Support For issues or questions: 1. Check the examples in `examples.py` 2. Run `python test_server.py` to verify functionality 3. Review server logs for error messages