Changelog

A Tools menu, workspace roles, Schema Compare, Settings Diff, and TTL inventory.

• Tools sidebar: SQL Console, Explorer, Explain, Advisor, Chart Builder, Schema Compare, and Settings Diff.
• Workspace roles (Full, DBA, Engineer, SRE). Hide pages from the sidebar or Settings.
• Schema Compare diffs CREATE TABLE across hosts or replica nodes. Copy-only plan.
• Settings Diff compares system.settings and merge_tree_settings. All matched when nothing differs.
• TTL and partition health lists every MergeTree table, including those without TTL.
• What's new dialog next to Settings.
• Switched from Cloud seats to self-hosted host-count licenses. No key in the binary.

Guest AI caps, simpler alerts, and pinned favorites you can drag.

• Cap and track guest AI usage on Cloud.
• Pick models dynamically, including AnyRouter presets and a custom model field.
• Alert settings use templates and presets instead of a wall of forms.
• See recently completed merges when none are running.
• Drag to reorder pinned favorites in the sidebar.
• Open Settings from the icon next to Sign in.

A denser fleet view, a live MCP playground, and alert channels that start from what you already configured.

• Alert channels show as a configured-first card grid.
• The query-activity heatmap picks a month window that fits the screen.
• Table Overview in Explorer is denser and engine-aware, with SQL highlighting for DDL.
• Fleet summary strip with richer host metrics and sparklines.
• MCP Playground talks to a live client and covers the 2026-07-28 spec gaps.

Interactive cluster topology and smarter model picking for the agent.

• Drag and inspect cluster topology with glass glyphs.
• Auto-pick AnyRouter models by real usage, not a static list.
• Surface why parts were postponed on 26.2+.
• Keep the Cloudflare Worker under the free 3 MiB limit.

The TanStack Start release — Cloud, Postgres, alerting, and AI on a new runtime.

• Dashboard runs on TanStack Start and Vite; Docker and Cloudflare Workers come from the same app.
• Hosted Cloud at dash.chmonitor.dev with a public demo and per-user connections.
• Monitor Postgres (beta) and PeerDB CDC next to your existing clusters.
• Alerting engine with templates, channels, and health reports.
• AI agent, Insights, advisors, and MCP at /api/mcp.
• Pluggable auth: none, Clerk, proxy, or trusted.
• Helm chart and one-click deploys; connection vars stay the same.

• Introduce outage escalation and error spike alerts for improved incident response.
• Add DAU/MAU tracking and weekly report generation to enhance user engagement analysis.
• Simplify the analytics page with distinct tabs for dashboard and CLI telemetry.
• Launch scheduled insights and cluster health report features with PDF export options.
• Revamp the agent page with a new design, including improved visuals and interactivity.
• Implement predictive parts-pressure early warning and memory CPU deep-dive charts.
• Enhance traffic monitoring with new performance and ingest speed charts.
• add centered logo to README (f74d754)

• Add CRON_SECRET and Kubernetes CronJob resources for improved cron job management.
• Rotate the overview hero on the landing page between two background screenshots for a dynamic look.
• Introduce a clickable findings feature on insights cards, opening detail dialogs with explanatory charts.
• Enable Postgres source (beta) by default on the cloud, enhancing database integration.
• Implement richer monitoring views for PeerDB, including snapshot, batch history, and fleet metrics.
• Add a compact hero section on the landing page with enhanced
• advisor: pick a query from quick examples or filtered history (#2608) (1714484)
• agent: page-context chip + dockable panel for the floating widget (#2587) (4284452)

Celebrate the arrival of version 0.2.14 of chmonitor, which rolled out over just 4 days with an impressive 105 commits! With contributions from an energized team of 4, including the tireless @github-actions[bot] who racked up 43 comments, this release showcases a brisk pace of 26.3 updates a day, with 72 commits made under the stars. Dive into the latest features that elevate your monitoring experience and address critical issues!

• Introduce the category-card homepage along with 6 new SEO blog posts to enhance discoverability.
• Pass the current dashboard page context to the chat agent for improved interactions.
• Implement fleet-wide rate limiting with a fallback mechanism to maintain performance.
• Add a percentile selector for average query duration to provide deeper insights.
• Launch a redesigned homepage featuring a live demo and changelog catalog.
• Add a near-ICP database comparison page to enrich user options.
• Introduce a zero
• agent: add anti-sycophancy instruction and eval case (#2459) (3e5afbe)

• Enable Cloudflare Workers Cache across workers for improved performance.
• Complete rebuild of the landing page with a new hero design, shadcn section, and refreshed screenshots.
• Add a per-processor duration breakdown to the query-stages chart for better insights.
• Introduce a SQL panel with syntax highlighting and a beautify toggle for easier query writing.
• Enrich query detail data with stack traces and upstream tables for comprehensive analysis.
• Enable an AI-generated dashboard and custom dashboard builder for tailored monitoring.
• Implement proactive weekly health reports and a query optimization advisor to guide users.
• Add a native Slack

• Add two-column "Add ClickHouse host" dialog with help panel.
• Implement conversation-level session stats aggregation in chat UI.
• Introduce a daily AI-quota indicator in chat UI.
• Enforce monthly AI USD budget and fix daily-usage accounting.
• Add incident system snapshot capture and endpoint for alert management.
• Include per-table summary in the new Overview tab of the explorer.
• Launch observability features with Sentry
• agent: Base UI chat thread + custom MCP server runtime (#2003) (756607a)

• Add an AI insights panel on the /insights page, merging plans into documentation for enhanced usability.
• Introduce browser notifications on by default, providing real-time updates without additional setup.
• Implement total-time sort, metric bars, and a cluster-load summary for expensive queries, improving monitoring visibility.
• Redesign insights settings with search capabilities and a live example to enhance user experience.
• Offer configurable AI insights generation options, including model selection and prompt style customization.
• agent: persist MCP server config to localStorage (#1833) (723f7b3)
• ai-insights: redesign panel + persist & auto-load insights + settings link (#1819) (bc9d633)
• alerts: browser notifications on by default (out-of-box, opt-out) (#1858) (6a21a28)

We're excited to announce the release of chmonitor v0.2.10! In just one day, our team delivered 8 impactful commits and 34 review comments. With contributions from 2 dedicated developers, including our star contributor, @github-actions[bot], who led the way with insightful feedback, this release is a testament to our commitment to enhancing your ClickHouse monitoring experience. Get ready for improved functionality and a more seamless installation process!

• Serve the chart repository via Cloudflare Pages at charts.chmonitor.dev for easier access and reliability.
• Ensure the /api/healthz endpoint is properly bounded and includes parameterized chart probes with dual-runtime guard.
• Make the chart-releaser idempotent by skipping existing releases during deployment.
• Fix the bootstrap process for the gh-pages branch on the first chart release to ensure successful initial deployments.
• Remove the unnecessary Pages-enable step that consistently caused
• agent: add AgentState as a conversation-history backend option (#1673) (9080602)
• agent: add scatter/radial/bar-list chart types to chat visualization (#1658) (a9b8a3f)
• agent: expand skill library with 9 new skills (#1655) (3accc86)