The quickstart takes you onto the platform in four steps: CLI, agent plugin, project, publishing. If Git, the terminal, and servers are new to you, continue with the For beginners path; if you develop every day, take the shorter For developers path. Both link to the same reference below, where each topic is described once.
The search at the top left covers headings and text of every page.
Quickstart
For beginners
What is the Application Platform?
The Application Platform takes care of repositories, pipelines, servers, and deployments so that you can concentrate on your application.
Terms you will come across
Repository, commit, pipeline, server, Docker, and workspace are terms you will meet constantly on the Application Platform, and each of them is explained briefly and in relation to the platform.
Create your first project
You register, create an organization, and create your first project in the wizard, which the platform generates with repositories in GitLab.
Bring in an existing project
A folder you built with Claude Code, Cursor, or Codex comes onto the Application Platform with one prompt, provided its tech stack is supported.
Set up your computer or work in the browser
You either work on your own computer, which the setup app prepares, or in a remote workspace in the cloud, and both paths lead to the same project folder.
Open and run your project
You open the project folder in the editor, start a terminal, run the application with ap run-local, and view it in the browser.
Work with an AI agent
Claude Code, Cursor, and Codex know the rules of your project and talk to the platform directly through the agent plugin.
Save changes and publish
With a commit and a push your changes land in GitLab, the pipeline brings them to the dev server, and you release production with a click.
When something does not work
Most typical stumbling blocks when getting started can be checked yourself, and if that is not enough, support helps.
What is GitLab?
GitLab is where the code of your projects lives and where the pipelines run, and day to day you only need a few of its functions.
What are pipelines?
A pipeline tests, versions, builds, and distributes your application after every push to main, and you start the last step to production yourself.
For developers
Overview in five minutes
Organizations, projects and the generated repositories, the path from push to production, and the two ways of working form the frame in which you work with the platform as a developer.
CLI and agent plugin
The ap CLI is your access to the platform from the terminal, and the agent plugin gives Cursor, Claude Code and Codex the same functions as MCP tools.
Create a new project
A new project is created in the wizard of the web interface or as JSON through the ap CLI and the MCP tools of the agent plugin.
Import an existing repository
A local folder becomes a platform project with repositories, pipelines and deployment, either through the agent or with ap platform projects import-local.
Local development
In the cloned workspace you start every application with ap run-local, use the runtimes from .tool-versions and follow the rules for generated and custom files, which apply to AI agents as well.
Git workflow and deployment
A push to main starts the pipeline, which derives the version from the commit messages, rolls out dev automatically and releases production through a manual job; an earlier state can be restored the same way.
Development environment
Working locally
The setup app prepares your machine and clones all repositories of a project into one shared folder where editors and AI agents see the whole project.
Remote workspace
A remote workspace is a prepared Ubuntu VM in the cloud where you work on your projects without installing anything on your own computer.
Starting applications
You start every application of a project with ap run-local in the root of its repository, and the command works the same locally on your machine and in the remote workspace.
Environment variables
The platform writes the generated env files, your own values belong in the custom files, and the deployment to dev and prod is configured in the gitops-configuration repository.
Editors and AI agents
Choose your editor
VS Code is the recommended starting point, and Cursor, Antigravity, JetBrains, Claude Code and Codex work with the same configuration from local-configuration.
Cursor, VS Code and Antigravity
The Application Platform extension and the configuration from local-configuration work the same way in VS Code, Cursor and Antigravity, and Cursor adds the agent plugin.
JetBrains
IntelliJ IDEA and other JetBrains IDEs work with platform projects, though with a leaner integration than VS Code.
Claude Code
Claude Code works in the terminal or as a desktop app, reads CLAUDE.md and .claude from local-configuration, and gets the agent plugin through its marketplace.
Codex
Codex runs as a CLI, desktop app or VS Code extension, reads AGENTS.md from local-configuration directly, and gets the agent plugin through its marketplace.
Your own rules for AI agents
In AGENTS.md and the editor rules from local-configuration you give all AI agents in your project your own working instructions, which apply to the whole team.
Connections and services
Connections
Under Connections you store access to Sentry, Firebase, Cloudflare, IONOS, Kubernetes, and email delivery once for your whole organization.
Docker Apps from the catalog
Under Docker Apps you run existing software such as WordPress, Nextcloud, or n8n on your own server without setting up a project of your own.
Email delivery
Backends in your projects send email through a mail account that you store for the organization, either as a Mailtrap connection or as a generic SMTP account.
Error tracking
Errors from apps, homepages, and backends land in your own Sentry account with stack trace and release version once you have connected it and enabled it per component.
Publishing
Servers
Backends, web apps, and websites run on a server that you connect to the platform in one of four ways.
Domain
To make your project reachable under a name of its own, you enter a domain and set the DNS records the platform shows you at your domain provider.
Apps (iOS and Android)
The pipeline builds and signs your app, distributes it to TestFlight and the Internal Track, and then waits for you to release it to the store.
Desktop applications
The pipeline builds installers for macOS, Windows, and Linux, signs them if you want, and uploads macOS apps to the Mac App Store when needed.
Web apps
A web app built with Flutter Web, React Native Web, or Next.js runs as a Docker container on your server, gets its dev address automatically, and uses your own domain in production.
Websites
Hugo and Astro sites consist of static files that the pipeline builds and delivers either to your Docker server or to an existing web space.
Backend
Your NestJS, FastAPI, or Laravel backend runs as a Docker container on your server, together with the MySQL database, the uploads, and the helper services that belong to it.
Backups
Easy Backup backs up the MySQL database and the uploads of your backend on a schedule or on demand and restores individual backups.
Operations
After the first deployment, the work shifts to spotting errors early, tracing releases, and going back to the last working state when needed.
Advanced
Libraries
A project can contain package repositories whose pipeline publishes your own packages and Docker images to the private GitLab registries of your organization.
Whitelabel apps
In whitelabel mode you run several brand variants from one app codebase, each with its own name, logo, domains, and store listings.
Architecture
GitLab pipelines build Docker images, a GitOps repository records which version runs where, and Ansible brings that state to servers that the platform has set up with Traefik, a firewall, and HTTPS.