Introduction
A read-only ClickHouse monitoring dashboard with an AI agent and MCP server — deploy on Docker, Kubernetes, or Cloudflare Workers.
chmonitor is a client-rendered ClickHouse monitoring dashboard. It reads from your ClickHouse system.* tables and renders query performance, storage, cluster health, and operational metrics directly in the browser. It does not SSR your ClickHouse data and writes nothing to the database except optional self-tracking events. An AI agent answers natural-language questions about your cluster, and an MCP server exposes the same monitoring tools to external AI clients.
Read-only by default
chmonitor does not manage schema, run DDL, or modify ClickHouse configuration. Optional action grants (kill query, optimize table) are off by default.
Who it's for
- Self-hosting teams who run ClickHouse on their own infrastructure and need an operational dashboard they can deploy and own.
- Operators who want to inspect running queries, track storage growth, monitor replication lag, or debug merge performance without writing ad-hoc queries by hand.
- Platform engineers who want to integrate ClickHouse monitoring into their existing auth and deployment stack (Docker, Kubernetes, Cloudflare Workers).
Pick your path
Getting started
Get a monitoring dashboard running in 5 minutes with Docker.
Install & configure
Deploy on Docker, Kubernetes, Cloudflare Workers, or Vercel.
Features
Understand what each dashboard page does and its feature id.
Authentication
Add login and fine-grained access control.
AI agent
Ask natural-language questions about your ClickHouse cluster.
Reference
Every environment variable, grouped by category.