Frequently Asked Questions about UBOS MCP Server

Q: What is an MCP Server?

A: An MCP (Model Context Protocol) Server acts as a bridge between a Large Language Model (LLM) and external data sources, standardizing how applications provide context to LLMs. It allows AI models to access databases, APIs, documents, and other resources.

Q: How does the MCP Server integrate with Claude?

A: The MCP Server provides a standardized interface for Claude to access external data. By configuring Claude to use the MCP endpoint, Claude can query the content of your PDFs and other data sources through the RAG system.

Q: What is a RAG system, and how does it relate to the MCP Server?

A: RAG stands for Retrieval-Augmented Generation. The MCP Server implements a RAG system, allowing Claude to retrieve relevant information from PDF files using Chroma as the vector database before generating a response. This enhances the accuracy and contextual relevance of the LLM’s output.

Q: What are the prerequisites for using the MCP Server?

A: You need Node.js (v14 or higher), npm (v6 or higher), Python 3.9+ with ChromaDB installed, and an OpenAI API key.

Q: How do I add PDF files to the MCP Server?

A: Place your PDF files in the data/pdfs directory of the project.

Q: How do I start the Chroma database server?

A: You can start the Chroma server using the ./start-chroma.sh script or manually using the command python3 -m chromadb.cli.cli run --path ./data/chroma_db.

Q: How do I ingest the PDF files into the vector store?

A: Run the command npm run ingest in a new terminal to process the PDFs, create embeddings, and store the vectors in the Chroma database.

Q: How do I start the MCP Server?

A: Start the MCP Server by running the command npm run dev in another terminal.

Q: How do I query the MCP Server?

A: You can query the server using a REST client by sending a POST request to http://localhost:3000/api/mcp/query with a JSON payload containing your query and the desired number of results (topK).

Q: What kind of data sources does the MCP Server support?

A: The MCP Server supports a variety of data sources, including PDF files, databases, APIs, and other external resources.

Q: Is the MCP Server scalable?

A: Yes, the MCP Server is designed to handle high volumes of requests and can be scaled to meet the demands of enterprise-level applications.

Q: How does the MCP Server ensure secure data access?

A: The server provides secure data access controls, ensuring that LLMs can only access the information they are authorized to see.

Q: What is UBOS, and how does the MCP Server relate to it?

A: UBOS is a full-stack AI Agent Development Platform focused on bringing AI Agents to every business department. The MCP Server is a key component of the UBOS platform, facilitating seamless data integration and enhanced AI Agent capabilities.

Q: What license is the MCP Server released under?

A: The MCP Server is released under the MIT license.

Q: Where can I find more information and support for the MCP Server?

A: Visit the UBOS website at https://ubos.tech for more information, documentation, and support resources.