Back

openable: run AI-generated apps without hassle

Lovable, Bolt, and v0 will happily generate a full React app for you. Then comes the cliff. The moment you export that project to GitHub and want to run it on your own machine, you’re suddenly expected to be a developer: install Node.js, pick the right version, run npm commands in a terminal, decode the errors when something breaks. The people these tools serve best are exactly the people that wall stops cold.

Openable is the missing link between AI code generation and local development. It’s a native desktop app we’re building at Wunderlabs that runs web projects straight from a git repository. Connect your GitHub account or paste a repo URL, click Start, and the project is running in seconds. No Node.js, no npm, no terminal. Everything is bundled inside the app.

Made for creators, not just developers.

What It Does

The demo above walks through the whole loop. Import a project generated by Lovable, watch Openable set everything up with visual feedback (so you understand what’s happening without needing to know what a dependency is), then open the running app in your browser. From there you can share it through a public URL, publish it, or jump straight into Cursor to keep iterating on the code.

Day-to-day it behaves like a project manager for AI-generated apps:

  • One-click projects. Paste a repo URL or browse your GitHub repos, including private ones via OAuth.
  • Zero prerequisites. Node.js, git, and the container runtime all ship inside the app.
  • Multi-project. Run several apps at once, each on its own port, with live logs streaming into the UI.
  • Git operations without git. Switch branches, pull the latest changes, and dependencies re-sync automatically.
  • Tunnel sharing. One click exposes a running project through a public Cloudflare Tunnel URL.
  • Cursor integration. Open any project in Cursor directly from the sidebar, which is exactly where AI-generated code wants to be edited.

How We Built It

The promise “no Node.js, no terminal” hides the actual engineering problem: how do you run an arbitrary dev server on a machine with no developer tools installed? Our answer is to bundle a lightweight container runtime per platform and hide it completely.

PlatformRuntime
macOS (Apple Silicon)vfkit on Virtualization.framework, with an Alpine Linux VM
macOS (Intel)QEMU with the same Alpine VM
WindowsWSL2 with an Alpine distribution
LinuxNative Podman, no VM needed

The VM image is a read-only Alpine base with Node.js pre-installed. At runtime the app never touches it directly. On macOS it creates an APFS copy-on-write clone, on Windows a fresh WSL2 distribution, so the base image stays pristine and a broken project can’t poison the next one.

The app itself is Tauri 2: a React 19 + TypeScript + Tailwind frontend over a Rust backend. When you start a project, Rust clones the repo via libgit2, mounts it into the container, and a small Rust agent inside the VM takes over, speaking gRPC (ttrpc) back to the host. The agent installs dependencies, starts the dev server, and streams logs out through Tauri events, which the frontend renders in real time. Port allocation, process lifecycle, and cleanup are all handled for you. Tunnel sharing runs through a Cloudflare Worker that coordinates the public URLs.

There’s no silver bullet in this design. Bundling a VM makes the app heavier than a thin wrapper would be, and maintaining four runtime paths is real ongoing work. The alternative was asking creators to install Docker, which for this audience is the same wall with a different logo. Zero prerequisites won.

Where It Stands

Openable is in beta and not yet released. macOS on Apple Silicon is the first platform, and the Intel macOS, Windows, and Linux runtime paths are already implemented and waiting their turn in the release pipeline. Join the waiting list to be the first to try it.

Frequently Asked Questions (FAQ)

Do I need to install Node.js or any developer tools?

No. Node.js, git, and the container runtime are bundled inside the app. If you can install an app, you can run your project.

Does it work with Lovable, Bolt, or v0 projects?

Yes. Openable is optimized for projects created by AI code generation platforms. Anything that exports to a GitHub repository can be imported and run.

What platforms does Openable support?

The beta targets macOS on Apple Silicon first. Intel macOS, Windows (via WSL2), and Linux (via Podman) runtimes are implemented and coming next.


Built by Vlad Temian and Marius Balaj at Wunderlabs.

Website · Join the beta

For the pixel cafe we built for AI agents at Wunderlabs, see cursouls. For a composable agent development environment that lives at the opposite end of the technical spectrum, see zag.

Stay curious ☕