v0.3 — What's New

v0.3 rebuilds the dashboard on TanStack Start, ships a pluggable authentication system, and delivers a batch of monitoring and UI improvements that accumulated during the v0.2 cycle.

New framework (TanStack Start)

The dashboard is rewritten on TanStack Start with Vite as the build tool. Every page is pre-rendered at build time so the initial shell loads from cache even before ClickHouse is contacted.

Static prerender for all 75+ dashboard pages.
TanStack Query replaces SWR for client-side data fetching.
Cloudflare Workers and Docker/Node builds produced from the same source.
NEXT_PUBLIC_* env vars renamed to VITE_*; old names still work as a fallback.

See Migrate to v0.3 for the one-step upgrade.

Pluggable authentication

A new server-side auth layer replaces the previous Clerk-only approach. Set CHM_AUTH_PROVIDER to choose how requests are authenticated:

Provider	Description
`none`	Default. Dashboard is open (no login required).
`clerk`	Clerk browser sessions — same as before.
`proxy`	Trust a reverse proxy: Cloudflare Access JWT or a trusted header + shared secret.

An always-on API key layer (CHM_API_KEY_SECRET) issues signed chm_ Bearer tokens for scripts and MCP clients, independent of the active provider.

See the Authentication reference for setup details and the decision flow.

Cluster topology visualization

The /cluster page is now a live topology diagram.

Shard and replica nodes are drawn with distinct shapes and the ClickHouse logo.
Physical cluster groupings are shown with labeled hull overlays.
Nodes are sized and colored by live metrics.
Nodes can be toggled on/off; the diagram adjusts height to the data.

Query page improvements

Running queries split into a live "Running" table and a "Recently completed" table with animated row transitions.
Slow, expensive, and failed query pages redesigned with expand panels, per-column filters, and highlighted columns.
EXPLAIN tree — the Explain page renders the query plan as an interactive tree instead of raw text. Five explain modes (Plan, Pipeline, AST, Syntax, Estimate).
Request Info dialog redesigned with formatted SQL and metadata badges.

Table/card layout for all data pages

Every data table now has a card grid view alongside the standard table view. Toggle between them per-page. Cards are the default on narrow screens.

Sitewide card toggle rolls out across all 54 query-config pages (tables, merges, replication, system, diagnostics, keeper, and more).
Rich expandable rows let you inspect SQL, metadata, and related actions inline without leaving the page.
Mobile layout uses SQL-hero cards so the most important column is always visible.

New monitoring views

Views added from ClickHouse system tables:

Kafka consumers — system.kafka_consumers at /kafka-consumers.
Part log — system.part_log redesigned with lifecycle charts, KPIs, and a filterable events table.
Query metric log — system.query_metric_log view at /query-metric-log.
User processes — system.user_processes with formatted badges and memory bars.
Moves — system.moves at /moves.
Dropped tables — system.dropped_tables at /dropped-tables.
Warnings — system.warnings at /warnings.
Replicated fetches — system.replicated_fetches at /replicated-fetches.
Replicated MergeTree settings — at /replicated-merge-tree-settings.

AI agent improvements

Conversation titles auto-generated from the first message.
Findings — the agent can now record and list persistent findings across sessions.
Workflow harness — dynamic workflow templates for multi-step analysis.
Multi-provider LLM config (LLM_API_KEY, LLM_API_BASE, LLM_MODEL) plus a redesigned chat UI with model selector.

MCP server improvements

Clerk OAuth login/consent flow for MCP clients — authenticate with an OAuth browser flow instead of a chm_ key.
In-process MCP route in the TanStack app; no separate worker required.

Deployment

Helm chart — a production-ready Helm chart is now vendored in the repo (deploy/helm/) alongside kustomize bases.
Kubernetes guide added to the docs.
Health alerting — a cron-based health sweep runs automatically and can post alerts.
CSV export — export any chart card's data directly to a CSV file.
Auto-refresh interval persisted to localStorage across reloads.
Time range persisted to localStorage and synced to the URL.

Breaking changes and upgrading

See Migrate to v0.3 for the complete upgrade steps. The only change most self-hosters need is a redeploy; the optional NEXT_PUBLIC_* → VITE_* rename is backward-compatible.