An AI Canvas in your browser

Organize and Edit text, images, and videos on a freeform canvas.
Run AI models on your content (using your own API keys) and then
combine it all into LLM-generated inline HTML pages.

View on GitHub Try Demo Learn More

Designed by a Human, Coded by a Robot

This project has been 100% coded by Claude Code, under the careful guidance
of rms80, an independent software developer.

What does it do?

Infinite Canvas Pan, zoom, and arrange without limits. Marquee selection, multi-select, and free-form positioning.
Text Blocks Editable text blocks with auto-sizing. Double-click to edit.
Images Import, Drag/Drop, or Paste from clipboard. Crop to fit (even GIFs!)
Videos Import or Drag/Drop video files with full playback controls. Crop, Trim, change speed, and remove audio.
Non-Destructive Image Cropping & Video Edits are non-destructive and can be modified or removed. Server stores original and edited versions.
AI Prompting Configurable prompt blocks that generate new Text, Image, and HTML content inline. Select canvas items as context.
HTML Pages AI-generated HTML rendered in sandboxed iframes, generated from selected canvas items & layout. Full web pages, live on your canvas.
HTML Publishing Export the inline HTML pages as HTML or Markdown files for publishing. Images are included in the export, referenced with local paths.
Multi-Scene Multiple Scenes managed with a Tabs UI. Pinned Tabs for quick access. Import & Export Scenes for backup or transfer.
Workspaces Scenes are grouped into Workspaces. Workspaces can be hidden, for your secret projects.

AI at your fingertips

Generate text, images, and web content directly on the canvas using leading AI models, using a Bring Your Own Key (BYOK) approach.

Text Generation

Create prompt blocks and generate text with Claude (Haiku, Sonnet, Opus) or Google Gemini (Flash, Pro). Select Text & Image items on the Canvas as context.

Image Generation

Generate and Edit images using Gemini Imagen models. Describe what you need and the image appears right on your Canvas.

HTML Generation

Generate and Edit full HTML pages from selected Canvas content. The AI understands the spatial layout of your workspace.

Run it your way

Three deployment modes to fit your workflow.

Online (S3)

Cloud storage with AWS S3 via the backend server. Full feature set with remote access. Run server locally or on the cloud (with basic password protection).

Local Disk

Store scenes on your local filesystem via the local server. Your data stays on your machine. No cloud services required.

Offline (Browser)

Fully client-side using IndexedDB. No backend needed. Configure your own API keys for AI features. No video support.

Built with modern tools

A robust, type-safe stack for both frontend and backend.

React 18 TypeScript Vite Konva Node.js Express Claude API Gemini API AWS S3 Sharp FFmpeg

Ready to try it?

gsworkspace is Open Source and free to use. Clone the repo and start building.

Get Started on GitHub
git clone https://github.com/rms80/gsworkspace.git