Frequently Asked Questions (FAQ) - MCP Server

Q: What is an MCP Server?

A: An MCP (Model Context Protocol) Server acts as a bridge, allowing AI models to access and interact with external data sources and tools. It standardizes how applications provide context to Large Language Models (LLMs), enabling more informed decision-making and automation.

Q: What are the key components of the MCP Server?

A: The MCP Server includes n8n (automation platform), Ollama (local LLM platform), Qdrant (vector store), Prometheus (monitoring), Grafana (visualization), Whisper (speech-to-text), Caddy (HTTPS/TLS), Supabase (database & auth), Flowise (AI agent builder), Open WebUI (ChatGPT-like interface), and SearXNG (privacy-focused search engine).

Q: What are the minimum system requirements for running the MCP Server?

A: You need an Ubuntu VPS (tested on 22.04 LTS), a domain name with DNS access, a minimum of 16GB RAM, 100GB+ storage, and Docker/Docker Compose installed.

Q: How do I install the MCP Server?

A: Connect to your VPS via SSH, install the required packages, configure the firewall, clone the repository, run the configuration script, and use the start_services.py script to start the AI stack.

Q: How do I access the different services after installation?

A: You can access the services via the provided URLs, such as https://n8n.kwintes.cloud for n8n, https://openwebui.kwintes.cloud for the Web UI, and so on.

Q: How do I update the MCP Server to the latest version?

A: Use the provided update_stack.sh script to pull the latest Docker images, apply configuration fixes, and restart all services.

Q: How do I back up the MCP Server data?

A: Use the backup_stack.sh script to back up all Docker volumes, configuration files, and secrets to a timestamped archive.

Q: What if I encounter Docker Compose issues during installation?

A: The script automatically detects and uses the correct Docker Compose command format for your system. If you’re manually running commands, use docker compose or docker-compose depending on your setup. If issues persist, install the standalone Docker Compose binary.

Q: How does the MCP Server ensure data privacy and security?

A: All AI models run locally on your VPS, preventing data from being sent to external services. Automatic HTTPS/TLS encryption, firewall rules, and secure secret management provide additional layers of security.

Q: How can I integrate the MCP Server with the UBOS Platform?

A: Use the UBOS Platform to orchestrate AI Agents running on your MCP Server, connect them to enterprise data, build custom AI Agents, and create sophisticated Multi-Agent Systems.

Q: What models are installed automatically?

A: Several models are automatically installed, including language models like gemma3:12b, granite3-guardian:8b, llama3.2-vision, and embedding models like granite-embedding:278m and nomic-embed-text:latest. These models are downloaded during the initial setup.

Q: How do I monitor the health and performance of my MCP Server?

A: Access Grafana to view dashboards or Prometheus to directly check the targets and create alerts. These tools provide detailed insights into service health and performance metrics.