sharingu
docs · v1

what is sharingu

multiplayer vibecoding for building real apps. anyone on the team can ship a change · the dev approves it · the system proves it.

sharingu is a live board on top of the apps your team is building. Anyone comments on the running app, from a desk or a phone. Agents build the change in gated, self-healing runs. Every change is verified on a real device before a PR lands. The dev reviews. The board repaints for everyone in seconds.

Nobody watches a terminal.

the loop

comment → cook → verify → self-heal → pr → live

A deterministic gate decides when something is done — the agent never declares it. If checks fail, the run heals itself and re-verifies, inside a hard spend cap. What lands in the PR has already been proven on a real device.

four surfaces

01atlas

the map. Every screen, route, and flow, regenerated from code on every push. Agents navigate it instead of re-exploring, and findings land on the screens where they happened.

02visionboard

design truth. Every element graded against your design system: expected token vs. observed. Drift is named, not argued about.

03the loop

self-healing runs. A deterministic gate decides done. Failures heal twice, then restart fresh, inside a hard spend cap.

04cockpit

one orchestrator for every app you're building. Track record, live context, comment-and-execute, phone-ready.

how teams run it

what it isn't

Not a test framework. Not another agent IDE. Not a way to skip review — it makes review the only thing the dev has to do.

honest scope

sharingu is for apps made of screens and flows: mobile and web. It's not for game engines or audio pipelines. Contributing from your phone works when your project's infrastructure is running.

where it stands

Functional V1, dogfooded daily on real production builds across React Native, Flutter, and web. Open-source release in prep.

get in

Tell us what you're building — we'll tell you honestly if sharingu fits.

reply with your stack