Section
Using CHouse UI
Every screen in the app, in the order of the app's own menu — what it shows, who can open it and how to work with it.
Home
Explorer
Databases & tablesTree-view schema inspection, create/drop databases, create/alter/drop tables across MergeTree engine families.SQL editorMonaco-powered editor with syntax highlighting, schema-aware completion, per-query execution statistics and history.Visual EXPLAINUnderstand a query's plan before you run it — EXPLAIN estimates, popout view and the debug dialog.Saved queries & historyPersist frequently used queries per connection, share them, and browse full query history.Upload & previewLoad CSV, TSV and JSON files into existing tables and sample data with pagination.ExportsDownload result sets and table samples as CSV, JSON or TSV from the workspace and explorer.AI AssistSchema-aware optimizer, debugger and chat with a pluggable provider list — OpenAI, Anthropic, Bedrock, Ollama and any OpenAI-compatible endpoint.
Data
How data is watchedThe collector, learned baselines, trust states and criticality behind every Data tab — and the grants it needs.OverviewOne screen for whether the data is right: trusted tables, critical coverage, open incidents with their root cause, pipelines and suggestions.Incidents, root cause & fixesCross-layer root cause and blast radius, investigation notebooks, fixes that run only after approval, Slack approvals and schema preflight.LineageHow data moves between tables, built from metadata and query_log without instrumentation — upstream sources and downstream blast radius.PipelinesKafka, RabbitMQ, NATS, S3Queue, AzureQueue, replication, views and writers in one status vocabulary, with samples and errors.Datasets & promisesObserved tables with their baselines, drift and usage, and promises — explicit checks evaluated on a schedule with incidents.CoverageHow well tables are protected, suggested promises for read-heavy tables, and cold data nobody reads.ContextTable descriptions, owners, canonical metrics, dbt import and plain-language watchers — for people and AI agents.Scheduled queriesCron and event-driven SQL jobs with macros, materialize destinations (also across a cluster), run history and replica-safe leases.Operational briefAI-generated summaries of a scheduled job's or promise's recent runs and incidents, on its detail page.
Monitoring
Monitoring overviewThe ten Monitoring tabs, the permission each one needs, and the system tables behind them.Live queriesRunning queries with CPU time and thread count, memory-pressure context and kill support.Query logsFive sub-views over system.query_log — Queries, Patterns, By table, By Redash and the duration/memory histogram.MetricsNine tabs of ClickHouse-native observability: overview, performance, storage, merges, errors, memory, CPU, ZooKeeper and network.Partssystem.part_log as stacked charts of merges, mutations, downloads and removals, plus a paginated event table.Schema advisorNullable-column and oversized-integer linter over system.parts_columns, ranked by on-disk bytes.ClusterMutations, replication queue, blocked-task indicators and per-replica lag from system.mutations and system.replicas.ErrorsA searchable viewer over system.errors and the crash log so recurring failures surface without ad-hoc SQL.PerformanceEach query shape against its own 14-day baseline, regressions lined up with the upgrades, DDL and setting changes around them.CapacityDisk forecasts per node, top growth, codec trials on samples and cost by consumer.UpgradesWhere your ClickHouse version stands, and what an upgrade changes for your workload: exact against a server on the target line, or from the release notes.Chouse AI in MonitoringOptimize query-log rows, fix system.errors rows and diagnose part-log rows without leaving the tab — read-only, before→after EXPLAIN.
Dashboards
Fleet
Doctor
Agents
AI GovernanceEvery MCP and token agent session, budget policies checked before queries run, health notices and the pause switch.AI activityWhat CHouse's own AI costs and sends out: tokens, estimated cost, ClickHouse queries and the data sent to LLM providers, per request.AI guardrailsRedact PII and secrets, cap results, stop runaway agents, refuse unsafe SQL and judge requests — in every AI run and MCP result, with shadow mode.Approvals & tiersHow AI changes are checked: undoable changes run with Undo, others ask once or need a second approver — 74 write tools in seven toolkits, and what AI never does.MemoryPreferences agents remember about you, and team knowledge about a connection that admins review before agents use it.Auto-triageAn agent investigates new incidents, fleet alerts, failed runs and Doctor findings on its own, within cooldowns and token budgets, and proposes fixes.
AI agents
AI agentsEvery AI feature and the chat run on an agent you can edit in AI Governance › Assistant: how features, agents, harnesses, skills and tools fit together.Editing agentsThe agent editor: prompt, tools, skills, subagents, model and tuning — and saving, duplicating, resetting and deleting agents.Prompt templatesFeature variables, conditional sections and skill includes in system prompts and task templates, and how the preview checks them.HarnessesHow DeepAgents runs an agent: which built-in tools it sees, the prompt suffix, the general-purpose subagent — and the three built-in harnesses.SkillsSKILL.md instructions and reference files that agents read on demand or pin into their prompt — anatomy, limits and writing your own.ChobiThe CHouse buddy above every dialog: Now briefing, Chat and Activity, nudges, watches, asking from anywhere, hand off and plain mode.Using the chatLive steps, Stop, files and skills with a message, changes with Undo, plans and approvals, memory notes, hand off and budgets in the chat.Chat agentsCHouse Assistant: one agent with toolkits, UI tools and read-only investigators — the agent picker, upgrading from CHouse Admin, and your own chat agents.Draft with AIDescribe an agent, harness or skill and get a complete, validated draft in the editor — every section filled or explained, redraft a section, refine with an instruction.Test consoleRun any AI feature or an unsaved draft as you and see the result, every tool call and the exact prompt — plus common errors.Agent evalsTest cases an agent must pass — expected tools and a judged rubric — run against the draft, with a threshold that gates saving it.Versions, upgrades & auditVersion history and restore, how CHouse upgrades built-in agents without overwriting yours, replicas, and the audit trail.AI feature catalogEvery AI feature with its permission, tool contexts, template variables and built-in agent — generated from the server.AI tool catalogEvery read-only tool agents can use, the DeepAgents built-in tools each harness keeps, and the built-in skills — generated from the server.
Admin
Admin overviewEvery Admin tab — users, roles, data access, connections, ClickHouse users and roles, audit, AI models, SSO, alerting — and who can open it.ClickHouse users & rolesCreate and manage real ClickHouse users and roles with native grants, from templates or by hand.AI modelsConnect AI providers, register provider models and publish deployments for Chouse AI — sixteen provider types, keys encrypted.AlertingChannels with health and what uses them, everything that sends alerts in one list, history with search, fleet threshold rules, and who can act from a channel.Chat appsSlack apps, Google Chat apps and Telegram bots: messages that update themselves and buttons to approve, acknowledge or snooze from chat.