MCP-Server
5
The project is a Model Context Protocol (MCP) server designed to enhance the capabilities of AI assistants by providing access to a customized document database. It overcomes AI knowledge cutoffs through document processing, embedding generation, and retrieval services.
❓ Troubleshooting
- Docker not found: Ensure Docker is installed and running. Check with
docker --version. - "Invalid reference format" error: This common error can occur for two reasons:
- Missing build step: You tried to run the MCP server without building the server image first. Always run
docker-compose build serverbefore attempting to run the server. - Mixing Docker and Docker Compose: The Windows batch file has been updated to use Docker Compose consistently. If you're still encountering this error, make sure you're using the latest version of the batch files from this repository.
- Missing build step: You tried to run the MCP server without building the server image first. Always run
- API key issues: Not to worry! You can use the free local embedding models without any API keys.
- Missing sentence-transformers package: If you select a free model, the system will automatically install the required package.
- Chroma database not found: Make sure you've run the pipeline to process your documents first.
- Connection issues: Verify the path in your MCP configuration points to the correct location of the run script.
- Windows path issues: If you encounter path problems on Windows, ensure paths use double backslashes (\) in the JSON configuration.
- Embedding model mismatch: The server automatically detects which model was used to create the database and uses the same model for retrieval.