Documentation
How Rigpaz works
A short tour of the editor, the in-browser runtime and the AI features — enough to get productive in a few minutes.
Getting started
Sign in with GitHub or Google — there is no password to create. Both providers resolve to the same account when they share an email address, so you can use whichever is convenient.
After signing in you land on the dashboard, where you create your first project. Nothing needs to be installed on your machine.
Open the dashboardProjects & templates
A playground is one project: its files, its chat history and its settings. Create one from a starter template — React, Next.js, Vue, Angular, Express or Hono — and it arrives with dependencies already declared.
You can also import an existing public repository. Paste its GitHub URL from the dashboard and Rigpaz fetches the tree, detects the framework and normalizes it into a playground.
Projects can be renamed, duplicated, starred and deleted from the dashboard table. Everything is scoped to your account.
The editor
Rigpaz uses Monaco, the same editor core that powers VS Code, so syntax highlighting, multi-cursor editing and the familiar keybindings all work as you'd expect.
The file tree on the left supports creating, renaming and deleting files and folders. Open files become tabs. Edits autosave as you work, so closing the tab mid-thought doesn't lose anything.
Runtime & preview
Your project runs in a WebContainer — a Node environment running inside the browser tab itself, not on a remote server. Dependency installs and dev servers execute locally, which is why there is no cold start to wait through.
The terminal panel is a real shell against that environment. The preview pane points at whatever dev server your project starts, and refreshes as you edit.
Because the runtime lives in the browser, everything resets when you close the tab. Your files persist to the database; the running process does not.
AI features
Chat lives in a side panel and runs on Groq. You can attach files from the current project to a message, which is how you give the model real context instead of describing your code to it. Conversations are stored per playground, so each project keeps its own thread.
Inline completion runs on a Mistral model wired directly into Monaco. Suggestions appear as ghost text while you type — press Tab to accept.
Shortcuts
- Open the command palette⌘K
- Accept an inline suggestionTab
- Save the current file⌘S
On Windows and Linux, use Ctrl wherever ⌘ is shown.
Known limits
- The in-browser runtime needs SharedArrayBuffer, so Safari support is limited. Use a Chromium browser or Firefox.
- Imports work with public GitHub repositories. Private repos are not supported yet.
- Native or platform-specific npm packages may not install, since the runtime is WebAssembly rather than a real OS.