• English

      • Quick Start

      Previously, AI Agents had difficulty understanding the NocoBase ecosystem — there were no standard tools for them to connect, read, and operate NocoBase data tables, pages, workflows, and more.

      Now, NocoBase fully supports AI Agent collaboration — you can use natural language to complete data modeling, page building, workflow orchestration, permission configuration, and even custom plugin development. Whether you use Claude Code, Codex, Cursor, or other AI tools, you can work with NocoBase through these two foundational components:

      • NocoBase CLI — The command-line tool (nb command), responsible for installing, managing, and connecting to NocoBase applications. AI Agents execute actual operations like creating data tables, building pages, and managing environments through CLI.
      • NocoBase Skills — Domain knowledge packages that enable AI Agents to understand NocoBase's configuration system and development conventions. They cover data modeling, UI configuration, workflow management, permission configuration, plugin development, and more. NocoBase CLI automatically installs Skills during initialization (nb init), so no manual installation is needed.

      Once you install CLI and complete initialization, your AI Agent can directly operate your NocoBase application.

      What Is NocoBase CLI

      NocoBase CLI (nb) is a command-line tool for initializing, connecting to, and managing NocoBase applications in your local workspace. It helps you prepare NocoBase applications, save CLI env configurations, and provides daily management commands such as start, stop, view logs, upgrade, and clean up — enabling AI Agents to connect and use NocoBase.

      CLI supports two common initialization methods:

      • Connect to an existing NocoBase application so that AI Agents can use it directly.
      • Install a new NocoBase application via Docker, npm, or Git, and save it as a CLI env.

      Prerequisites

      Before installation, make sure the following environment requirements are met:

      • Node.js 22
      • Yarn 1.x

      If you already have a NocoBase instance, it needs to be version >= 2.1.0-alpha.22.

      If performing a fresh NocoBase installation, also confirm the following:

      • A database is ready (PostgreSQL, MySQL, or MariaDB), or use the built-in Docker database service
      • Docker (if deploying via Docker)
      • Git (if installing from source)

      Installation

      Manual Installation
      Install with AI Agent

      Step 1: Install CLI

      npm install -g @nocobase/cli@beta
nb --version // Confirm successful installation

      Step 2: Create a Working Directory

      You can customize the working directory name, but avoid spaces or special characters to prevent issues later. All subsequent CLI commands need to be run in this directory.

      mkdir my-nocobase && cd my-nocobase

      Step 3: Initialize

      nb init --ui

      The browser will automatically open a visual configuration page, guiding you through 5 steps:

      1. Start Configuration — Set the application name (also used as the CLI environment name), choose between a fresh installation or connecting to an existing application
      2. Create Application — Choose the installation method (Docker / npm / Git), set port and other basic configurations
      3. Download Application Files — Select source and version, download NocoBase source code or image
      4. Configure Database — Use the built-in database or configure an external database (PostgreSQL, MySQL, MariaDB)
      5. Create Admin Account — Set the admin email and password

      If you choose to connect to an existing application, steps 2-5 are skipped — just fill in the API address and select the authentication method (OAuth is recommended for quick setup).

      nocobase cli visual wizard

      After a fresh installation is complete, open your browser and visit http://localhost:13000 (default port), then log in with the admin account you just set up.

      (If you prefer working in the command line, you can also run nb init to enter the interactive wizard, or nb init --env=app -y to use default configuration with one command.)

      Step 4: (Optional) Connect AI Agent to NocoBase

      If you want an AI Agent to participate in operating NocoBase, you just need to:

      1. Enter the working directory that CLI has already initialized (e.g., my-nocobase)
      2. Start a new or restart the AI Agent session in this directory (if using a desktop GUI AI Agent tool, add the CLI-initialized directory as the working directory), then you can let the AI Agent start operating NocoBase.

      What's Next

      Manage Application Environments

      If you need to manage existing NocoBase instances (such as test or production environments), or switch between multiple environments, see Environment Management — covering environment checks, adding environments, upgrades, and troubleshooting.

      Start AI Building

      Describe your requirements in natural language, and AI helps you complete data modeling, page building, and workflow orchestration — from a single sentence to a complete business system. See AI Building Quick Start.

      Continue with AI Plugin Development

      If you need to develop custom plugins, AI can help too — from scaffolding generation to component code and business logic, covering the full plugin development process. See AI Plugin Development.

      Use AI Employees

      Collaborate with AI employees directly in the NocoBase interface — data analysis, report generation, translation, decision support, without leaving your business system. See AI Employees.

      NocoBase CLI Command Reference

      If you already have a NocoBase application running, or have successfully completed initialization and want to continue developing, debugging, or managing your application, see NocoBase CLI Command Reference for full parameter documentation of all nb commands.