Frequently Asked Questions (FAQ) about UBOS MCP Server

Q: What is the Keboola MCP Server?

A: The Keboola MCP (Model Context Protocol) Server is an open-source bridge that connects your Keboola project with modern AI tools. It allows AI agents and MCP clients to access and interact with your Keboola data, transformations, SQL queries, and job triggers without needing complex code.

Q: What MCP clients are compatible with the Keboola MCP Server?

A: The Keboola MCP Server supports a range of MCP clients, including Claude (Desktop & Web), Cursor, Windsurf, Zed, Replit, Codeium, Sourcegraph, and custom MCP clients using HTTP+SSE or stdio.

Q: What are the key features of the Keboola MCP Server?

A: Key features include storage access (querying tables, managing descriptions), component management (listing extractors, writers, etc.), SQL transformation creation with natural language, job execution and detail retrieval, and metadata searching and updating.

Q: What prerequisites are required to run the Keboola MCP Server?

A: You need Python 3.10+ installed, access to a Keboola project with admin rights, your preferred MCP client, and uv installed to automatically download and run the server. Also, you will need KBC_STORAGE_TOKEN, KBC_WORKSPACE_SCHEMA.

Q: How do I obtain the KBC_STORAGE_TOKEN?

A: The KBC_STORAGE_TOKEN is your authentication token for Keboola. You can create and manage Storage API tokens in your Keboola project settings. Refer to the official Keboola documentation for instructions.

Q: How do I find my KBC_WORKSPACE_SCHEMA?

A: The KBC_WORKSPACE_SCHEMA identifies your workspace in Keboola and is required for SQL queries. Follow the Keboola guide to get your KBC_WORKSPACE_SCHEMA.

Q: What if my Keboola project uses BigQuery?

A: If your Keboola project uses BigQuery, you need to set the GOOGLE_APPLICATION_CREDENTIALS environment variable in addition to KBC_STORAGE_TOKEN and KBC_WORKSPACE_SCHEMA. This variable should point to the full path of your BigQuery credentials JSON file.

Q: How do I run the Keboola MCP Server?

A: There are four ways to run the server: Integrated Mode (recommended, client starts it), Local Development Mode (for developers), Manual CLI Mode (for testing), and using Docker.

Q: Which deployment option should I choose?

A: Integrated Mode is recommended for most users. If you are developing the MCP server locally, use Local Development Mode. Manual CLI Mode is for testing or debugging, and Docker is suitable for containerized deployments.

Q: Do I need to start the server manually if I’m using Claude or Cursor?

A: No, if you are using Claude or Cursor, configure the MCP settings in the app, and the client will automatically start the server when needed.

Q: What are some examples of queries I can use with the Keboola MCP Server?

A: You can explore data by asking “What tables contain customer information?”, analyze data with “Run a query to find the top 10 customers by revenue”, and manage data pipelines with “Create a SQL transformation that joins customer and order tables”.

Q: Where can I get help or report issues with the Keboola MCP Server?

A: The primary way to get help, report bugs, or request features is by opening an issue on GitHub. The development team actively monitors issues and will respond as quickly as possible.