Section
Guides
Step-by-step walk-throughs of the jobs people do most — from a stale table to a fix, from a new team to its first agent.
Investigate a stale tableFrom "the dashboard shows yesterday" to the root cause, the people affected and a fix.Approve and roll back a fixSet up a remediation credential, then propose, approve, run, verify and roll back one fix.Let Chobi make a changeSet up AI changes once, then meet all three tiers: a change with Undo, one that asks first, and one that needs a second person.Approve from SlackConnect a Slack channel as a chat app, send approvals and alerts there, and act on them with buttons that run as you.Build a dashboard on the canvasFrom a template to a tidy canvas, then what changed, whether it is normal, why it spiked, and sharing what you saw.Watch a Kafka pipelineMake sure a Kafka consumer's failures — including silent retries — get noticed.Set up your first promiseA freshness promise with an alert, from choosing the table to validating the generated SQL.Customize an AI featureTeach Optimize your team's SQL conventions with a skill and a copied agent, test it, switch over and roll back.Build a chat agentA team chat agent with its own data skill and tools, gated by permission, optionally routed to by Auto.Connect an AI agentTurn MCP on, give Claude Code a narrow token or let an agent sign in, and watch and limit what it does.Use the CLI in CIA scoped token, JSON output and exit codes for reports and data checks in a pipeline.Plan a ClickHouse upgradeChoose a target, check what it changes for your workload, fix the queries the new settings break and watch the rollout.Give a team least-privilege accessA data access policy and a custom role that let a team read one database and nothing else.Link an account to a ClickHouse userMake an account's statements run as its own ClickHouse user on a personal connection: the link, the grant, and what suspends it.