Keboola Docs

CLI

kbagent, the Keboola command-line interface — an AI-friendly CLI to manage projects, configs, jobs, storage, flows, and dev branches across your whole organization, driven by you or your coding agent.

kbagent is Keboola’s command-line interface: one tool to manage all your Keboola projects from the terminal. It’s built to be equally comfortable in your hands and in an AI coding agent’s — every command has structured JSON output and the whole thing sits behind a safety firewall you can tighten before handing it over.

What you can do with it

Manage projects across your whole org — connect one project, several, or an entire organization, across US/EU and AWS/Azure/GCP stacks, and operate on them together.
Browse and edit configurations, components, buckets, tables, and files — the same objects as the UI, scriptable.
Run and monitor jobs — trigger a component, wait for it, list failures across every project, terminate a stuck one.
Build and schedule flows, create SQL workspaces to debug transformations, and manage development branches.
Version your project with GitOps — pull configurations to local files, diff them, and push changes back.
Search everything — find a config, table, or flow by name or content across all connected projects.
Hand work to an AI agent safely — a read-only sandbox and a permission firewall let an agent operate Keboola without risk.

In one minute

# install (macOS/Linux; see Get started for Windows)
curl -LsSf https://raw.githubusercontent.com/keboola/cli/main/install.sh | sh
# connect a project
kbagent project add --project prod --url https://connection.keboola.com --token YOUR_TOKEN
# explore
kbagent doctor
kbagent job list --limit 5
kbagent search "customer_id"

Everything the UI does — and a lot it doesn’t — is a command away, scriptable, and safe to hand to an agent.

Why a CLI (and not just the UI)?

The UI is great for one project, one change at a time. kbagent is for the things the UI makes hard: doing the same thing across many projects at once, putting your configuration under version control, wiring Keboola into CI/CD, and letting an AI agent do the work while you keep approval of anything destructive. It talks only to your Keboola stacks, stores its connections locally, and never needs a browser.

This section

The pages read in order, from first run to deep reference:

Get started — install (macOS / Linux / Windows), connect a project, run your first commands.
How it works — the connection model, multi-project, dev branches, GitOps sync, and the safety firewall.
How-to guides — task recipes: onboard an org, dev-branch workflow, GitOps sync, audits, CI/CD tokens, encryption.
Use with AI agents — the Claude Code plugin, kbagent context, and sandboxing an agent.
Command reference — every command group, global flags, JSON output, and error codes.
Web UI — the optional local browser dashboard.

How it relates to the other tools

Tool	Use it for
kbagent CLI	Scripting and multi-project operations from your terminal; giving an agent safe, sandboxed control.
MCP server	Letting an AI client call Keboola tools directly over MCP.
AI Kit	Coding-assistant plugins for building Keboola components and apps.
Kai	The in-product AI assistant.

Next: Get started with kbagent →