7.7K ·GitHub
What is this?
An open-source infinite canvas workbench for AI creation: it puts canvas orchestration, AI image generation, reference-image editing, a chat assistant, a prompt library and asset management in one interface.
An open-source infinite canvas workbench for AI creation: it puts canvas orchestration, AI image generation, reference-image editing, a chat assistant, a prompt library and asset management in one interface.Is it a browser app, or something I host on a server?Both halves: the canvas UI runs in your browser and calls your configured OpenAI-compatible endpoints directly from the browser, while you serve the app itself via a local dev build or Docker (default port 3000).
Both halves: the canvas UI runs in your browser and calls your configured OpenAI-compatible endpoints directly from the browser, while you serve the app itself via a local dev build or Docker (default port 3000).How do I get it running?Run it with Docker or as a local dev build, open it on port 3000, then go to the top-right settings and fill in your own OpenAI-compatible Base URL and API Key.
Run it with Docker or as a local dev build, open it on port 3000, then go to the top-right settings and fill in your own OpenAI-compatible Base URL and API Key.Do I need to bring my own AI API, or does it include one?It includes no model access — you supply any OpenAI-compatible Base URL and key. If you lack an image API, the README points to chatgpt2api as a free image-generation option; relays like grok2api, flow2api and newapi also work.
It includes no model access — you supply any OpenAI-compatible Base URL and key. If you lack an image API, the README points to chatgpt2api as a free image-generation option; relays like grok2api, flow2api and newapi also work.Is it safe to build up long-term work on it right now?Not yet guaranteed: the project is under active development, historical data compatibility is not assured, and local storage formats may change outright — the author suggests forking if you need a stable branch.
Not yet guaranteed: the project is under active development, historical data compatibility is not assured, and local storage formats may change outright — the author suggests forking if you need a stable branch.Where do my API keys and generated work end up?In your browser by default: the API key, Base URL, canvases, assets and generation records are all saved to local browser storage, and the prompt library is fetched by the frontend and cached in IndexedDB.
In your browser by default: the API key, Base URL, canvases, assets and generation records are all saved to local browser storage, and the prompt library is fetched by the frontend and cached in IndexedDB.What kinds of AI generation does it support?Text-to-image, image-to-image, reference-image editing, text Q&A, and audio and video generation — all through the OpenAI-compatible endpoints you configure, with customizable call scripts if your provider's API differs from the default.
Text-to-image, image-to-image, reference-image editing, text Q&A, and audio and video generation — all through the OpenAI-compatible endpoints you configure, with customizable call scripts if your provider's API differs from the default.Can coding agents like Claude Code or Codex operate the canvas?Yes. A local Canvas Agent connects Codex or Claude Code over MCP so an agent can manipulate the current canvas, and there is a Codex app plugin that auto-registers MCP and launches the local agent on install.
Yes. A local Canvas Agent connects Codex or Claude Code over MCP so an agent can manipulate the current canvas, and there is a Codex app plugin that auto-registers MCP and launches the local agent on install.What can the canvas itself do?Multiple canvas projects with node dragging and zooming, node connections, a minimap, undo/redo, and import/export.
Multiple canvas projects with node dragging and zooming, node connections, a minimap, undo/redo, and import/export.Is there a built-in prompt library?Yes — it bundles 7 open-source prompt sources, supports custom sources in a standard JSON format, and loads and caches them in IndexedDB directly from the browser.
Yes — it bundles 7 open-source prompt sources, supports custom sources in a standard JSON format, and loads and caches them in IndexedDB directly from the browser.What are similar projects?Open-Generative-AI, dramaclaw, awesome-gpt-image-2-API-and-Prompts, gpt_image_playground, Toonflow-app, banana-slides, GPT-Image2-Skill, printfilm
Open-Generative-AI, dramaclaw, awesome-gpt-image-2-API-and-Prompts, gpt_image_playground, Toonflow-app, banana-slides, GPT-Image2-Skill, printfilmWhich topics is it in?localstorage, prompt-library, creative-tools, plugins, image-editing, text-to-image, canvas, video-generation, openai-compatible, image-generation, generative-ai, docker-compose
localstorage, prompt-library, creative-tools, plugins, image-editing, text-to-image, canvas, video-generation, openai-compatible, image-generation, generative-ai, docker-compose