A GUI sketchpad for 'vibe coding' just hit 6,000 GitHub stars – but stars aren't proof of anything
m3e-canvas lets you doodle an app screen and spit out a prompt for Claude Code or Cursor, and it's racked up thousands of stars in under a fortnight – here's what that number actually tells you.
What it actually is
M3E Canvas is a browser-based sketching tool that lets developers rough out app screens using Google’s Material 3 Expressive design language, link the screens together so you can click through a rough prototype, and then export the whole thing as a text prompt. That prompt is meant to be handed to an AI coding assistant – the project page name-checks Claude Code, Codex, Gemini CLI and Cursor – which then generates the actual working app.
In plain terms: it’s a wireframing tool with a “copy prompt” button bolted on the end, aimed at the growing crowd of developers who describe apps to an AI rather than typing every line themselves. That workflow has picked up the label “vibe coding” over the past year, and this tool is built specifically to feed it.
The repository is open source under an MIT licence, written in TypeScript, and has had 143 commits. It ships localised documentation in Japanese, Chinese and Korean, suggesting the developer or early contributors were courting an international audience from fairly early on.
The number everyone’s talking about
The headline figure is star count: nearly 6,000 stars and 586 forks in under two weeks, according to GitHub’s own count at time of writing. That’s a genuinely fast climb by GitHub standards – most repositories never get near that in years, let alone days.
But a star count measures attention, not quality. GitHub stars can be driven by a well-timed post on Hacker News or X, by bot farms, or by developers bookmarking something they’ll never actually open again. None of that is visible from the star tally alone. Nothing in the repository’s public listing confirms how the growth happened, and the tool’s own maintainers don’t publish that data either.
The seven open issues and five pull requests suggest a project still being actively poked at rather than one that’s settled and mature – which is normal for a two-week-old viral repo, but worth noting if you’re expecting something polished.
So who is this actually for
This is a developer tool, not a consumer app. If you don’t already use an AI coding assistant, M3E Canvas does nothing for you – there’s no standalone product here, just a bridge between a sketch and a prompt. Even for developers, the tool doesn’t write any code itself; it hands off to whichever AI assistant you plug in, and the quality of what comes out depends entirely on that assistant, not on M3E Canvas.
Google’s Material 3 Expressive is a genuine, documented design system, so the sketching side has a real reference point. What’s unverified is how faithfully a hand-drawn sketch translates into a usable prompt, and how good the resulting AI-generated app actually is once you run it. The project’s own demo video shows a recipe app being sketched, restyled and rebuilt – but a demo is a demo, not an independent test.
The takeaway
M3E Canvas is a real, working repository with real commit history, not vapourware — but its rocketing star count is a measure of buzz, not of proven usefulness. If you’re already deep in the AI-coding workflow it’s worth a look; if you’re not, this is a niche tool for a niche crowd, and the numbers alone shouldn’t be mistaken for a verdict on how well it works.