what is canvashub fr?

CanvasHub is a local-first console for generating and managing AI-created images. It runs on your own computer through a small Python server that serves both a desktop-style web interface and the JSON APIs behind it, and it can also be delivered as a Telegram Mini App so people can generate images from within Telegram. Generated images are saved locally, and the Telegram integration syncs those saved files. The tool is built around node-based image generation, meaning you connect pieces like prompts, image imports, and generation steps rather than filling in a single form. Two image generation paths are supported: one called Image2, which needs a Codex login or account credentials imported through an account-pool system, and one called Banana generation, which needs a separate third-party API. Beyond generation itself, CanvasHub includes prompt management tools and an optional feature that syncs example prompts and images from several public, curated prompt repositories into your local library. Security is opt-in by design. The server only listens on your own machine by default, and turning on public mode, which is needed to expose the app to the internet or use the Telegram Mini App remotely, requires you to explicitly set an access password in the settings. The README is upfront that this password gate is basic protection, not a full multi-user permission system, so anyone deploying it publicly is expected to add their own HTTPS, reverse proxy, and firewall rules. Setup uses a standard Python virtual environment and a requirements file, with one-click start scripts provided for Windows, macOS, and Linux. A separate desktop app build is also available through GitHub Releases. The project's own code is released under the MIT License, though some bundled third-party components carry their own separate licenses.