Hover Cloud: a Playwright CI dashboard with a heal queue

Your Playwright suite already runs in GitHub Actions. What it doesn't do on its own is tell you which specs are getting flaky, which failures are real bugs versus moved selectors, and what your suite still doesn't cover. Hover Cloud reads the runs you already produce and answers those questions. It runs no browsers of its own.

What Hover Cloud does

Sign in with GitHub and connect a repo. Hover installs the app, writes the CI workflow, and sets the reporting secrets. From then on every run reports back, and the dashboard turns those runs into trends, a triage queue, and alerts. It is free while in early access.

A Playwright CI dashboard

The dashboard aggregates each run across your projects: pass rate, run history, and a flakiness score that ranks your shakiest specs by how often they flip between pass and fail. Instead of reading raw Actions logs, you see which spec is costing you confidence and how long it has been doing it.

The Heal Queue: drift vs bug

A red run is one of two things. Either the UI moved and the test needs a new selector (drift), or the app actually broke (a bug). Hover Cloud classifies each failure and puts it in a Heal Queue. Drift gets a heal command routed to your editor, where you fix it locally and review the diff. A real bug stays flagged. Nothing in the cloud rewrites your spec on its own.

Business map and knowledge

Cloud reads the business map and remembered rules your repo already commits under .hover/. You get a coverage graph of your flows (which areas are tested, which are gaps) and a timeline of the rules Hover has learned about your app, visible to the whole team.

Alerts, PR summaries, environments

Regression alerts reach you by email or Slack when a green suite starts failing. Pull requests get a coverage note and a plain-language failure summary. Environments let one suite target staging, production, or a preview deploy, with test-account secrets kept out of the spec.

It runs no browsers

Execution stays in your CI on your own Actions minutes. Cloud only reads results. Your specs stay plain @playwright/test in your repo, and you could drop Hover tomorrow with every test still running. That is the self-heal loop without a model in CI, lifted to your whole team.

Get started

Author specs with the Hover MCP, run them in CI, then connect the repo at cloud.gethover.dev. The Cloud docs walk through connecting a repo, environments, and the heal loop.