Mcp Clickhouse
Run read-only ClickHouse analytics, inspect tables, and check cluster health from your coding agent without opening a SQL client.
Overview
mcp-clickhouse is a MCP server for the Grow phase that gives AI agents read-only ClickHouse analytics queries, table inspection, size metrics, and health checks.
What is this MCP server?
- Read-only analytics queries against ClickHouse
- Table inspection, sizes, and health-oriented tooling
- No write path—safer for production warehouse exploration
- stdio npm package @infoinlet/mcp-clickhouse v0.1.1
- From infoinlet-marketplace/services/mcp-clickhouse
- Server version 0.1.1
- npm package @infoinlet/mcp-clickhouse with stdio transport
- Read-only scope: analytics queries, table inspection, sizes, health
What problem does it solve?
Jumping between your agent and a ClickHouse client slows answers when you need table stats, ad-hoc SQL, or warehouse health during growth debugging.
Who is it for?
Solo builders running ClickHouse for product analytics who want Claude Code or Cursor to explore data and schemas safely.
Skip if: Greenfield projects without ClickHouse, or teams that need INSERT/DDL and migrations from the agent.
What do I get? / Deliverables
With MCP registered and ClickHouse credentials configured, agents return query results and metadata from your warehouse through controlled read-only tools.
- Agent-callable read-only ClickHouse query and inspection tools
- Faster answers on schemas, sizes, and health without leaving the IDE
- Safer analytics exploration without write-capable SQL from the model
Recommended MCP Servers
Journey fit
Product metrics and warehouse inspection sit in Grow analytics—where solo builders validate funnels and usage after ship. ClickHouse is the analytics engine; this MCP keeps queries and metadata inspection inside agent sessions for growth and ops questions.
How it compares
Read-only ClickHouse MCP integration, not a full BI dashboard or ETL skill.
Common Questions / FAQ
Who is mcp-clickhouse for?
Developers and founders using ClickHouse for analytics who want MCP-enabled agents to run exploratory SQL and inspect tables without a separate SQL IDE.
When should I use mcp-clickhouse?
Use it in Grow or Operate when investigating retention, funnel queries, table bloat, or warehouse health after you have shipped and collect events in ClickHouse.
How do I add mcp-clickhouse to my agent?
Install @infoinlet/mcp-clickhouse via npm, configure stdio MCP in your agent with ClickHouse connection settings from the marketplace repo, and restart the client so read-only tools are available.