MCP: ClickHouse
by ClickHouse
ClickHouse's official MCP server — run analytical SQL, explore schemas, and query local files with chDB.
Apache's official read-only MCP server for Doris — eight tool domains and 55 progressively disclosed children covering catalog, query, cluster, lineage and lakehouse.
Set up this MCP server in your favorite AI agent — copy a ready-made configuration below.
The Apache Doris MCP Server exposes the real-time analytics database to agents over the MCP 2026-07-28 specification, and its 1.0 release is a deliberate rebuild of the tool surface. Instead of dozens of flat tools competing for context, it presents eight stable domains — doris_catalog, doris_query, doris_cluster, doris_pipeline, doris_search, doris_governance, doris_lakehouse and doris_semantic — each of which is called with {} to discover the authorised children and their exact schemas, then called again with a child_tool, arguments and the returned manifest_version. Fifty-five children sit underneath, covering everything from table context and query profiling through node metrics, compaction and workload groups to ingestion freshness, text/vector/hybrid search, data lineage, audit and external lakehouse catalogs. Hosts that cannot do progressive disclosure can set MCP_TOOL_EXPOSURE_MODE=flat.
The catalogue is read-only by design — doris_admin is reserved and not registered — and authorisation is explicit, with static tokens, JWT, external OAuth/OIDC and Doris-backed OAuth all supported, plus per-request RBAC against the Doris route.
Requires Python 3.12+, Apache Doris 2.0.0+ and network access to the FE MySQL endpoint (usually port 9030). Both stdio and Streamable HTTP are GA; a legacy HTTP adapter (ENABLE_LEGACY_HTTP_ADAPTER=true, /mcp/legacy) covers hosts still on the 2025-06-18 handshake. Apache 2.0 licensed.
by ClickHouse
ClickHouse's official MCP server — run analytical SQL, explore schemas, and query local files with chDB.
by Databricks
Databricks-hosted MCP endpoints that expose Unity Catalog functions, Genie spaces, Vector Search indexes and Databricks SQL to agents, with UC permissions enforced.
Metabase's built-in MCP server — search, query and visualise your BI data through the semantic layer, scoped to the connecting person's existing Metabase permissions.
by Keboola
Opens a Keboola project to an agent: query storage tables, write SQL transformations, configure extractors and writers, trigger jobs and build flows.