Tests written the way you describe them
End-to-end testing is where good intentions go to die. Everyone agrees Playwright coverage matters, and then the backlog of untested flows keeps growing because writing and maintaining the tests is slow, fiddly work. Litmus Check attacks that directly: you describe what a test should do in plain language, and it produces the Playwright code, with a choice of selectors for each element and space to drop in custom scripts where you need them.
When a test breaks, you don't start from a red stack trace. An agent triages the failed run, reproduces it, and tells you what actually went wrong. Teams using Litmus Check write and maintain Playwright suites roughly five times faster than doing it by hand.
Open source, top to bottom
We built Litmus Check as a hosted product, then opened the whole thing. The engine, the UI, and the triage CLI are all public and free to self-host. Nothing important is held back behind a paywall. The project lives under github.com/litmus-check as three repositories:
- lc-server (Python). The QA engine that turns natural-language intent into Playwright tests.
- lc-frontend (TypeScript, Next.js). The self-hostable UI, deployable on Vercel or your own VMs.
- litmus-agent (Node.js). The CLI that triages failed tests from a JSON report.
Each repository ships with its own license. Read the code, file issues, send pull requests, or fork it and make it your own.
Who it's for
Opening the source changes who gets to use it. The people who benefit most tend to share a problem: real testing needs, and not enough time or headcount to meet them.
- Small teams. Get meaningful Playwright coverage without hiring a dedicated automation engineer to write it all.
- Manual QA moving to automation. Describe the checks you already run by hand and get code you can commit, instead of learning a framework from scratch first.
- Engineers who want control. Self-host the engine, read exactly how generation and triage work, and adapt it to your stack. No vendor lock-in, no per-seat gate on the core.
Why we opened it up
Testing tools work best when you can see inside them. A generated test you can't inspect is a liability, and a triage verdict you can't audit is just another black box. Open source removes both problems: you can read every line that produces your tests, change what doesn't fit, and keep running the whole system even if our roadmap goes somewhere you don't want to follow. If you want the managed version, litmuscheck.com still hosts it. If you'd rather own the whole stack, clone the repositories and go.