Keboola MCP Server is an open-source bridge between Keboola projects and modern AI tools. It turns Keboola features—such as storage access, SQL transformations, and job triggers—into callable tools for Claude, Cursor, Windsurf, LangChain, Amazon Q, and more. It supports remote hosted MCP Servers with OAuth authentication for multi-tenant Keboola stacks and can connect to AI assistants via remote SSE or HTTP transports. The server exposes Keboola capabilities as accessible tools for AI assistants, enabling data exposure, transformations, and job orchestration without glue code. Key capabilities include storage queries and metadata management, component management (extractors, writers, data apps, and transformation configurations), natural-language SQL transformations, and job execution data. It also provides workflow tooling (Flows and Orchestrator Flows), Keboola Streamlit Data Apps, and metadata search/update. The project supports development branches and multiple transports (stdio, streamable-http, SSE, http-compat). Deployment options include remote server usage, local development, manual CLI mode, and Docker. Security and credentials are managed via KBC_STORAGE_API_URL, KBC_STORAGE_TOKEN, KBC_WORKSPACE_SCHEMA, and KBC_BRANCH_ID, with per-request overrides available via headers.