Lark

Lark is a continuous testing service that runs end to end tests against a product's user interface, APIs, command line tools and mobile apps on a constant schedule. Tests are written in natural language rather than code, and the vendor says its agents can also cover SDKs and asynchronous workflows. It is aimed at engineering teams that want a test suite that does not break each time a dashboard is redesigned or an API changes. The core feature is self healing: when the product changes, Lark adapts the test and alerts the team to the change. The vendor compares this with Playwright, stating that Playwright tests are written in code, can be brittle and cover the front end only, while Lark tests cover the front end, back end APIs and more. Runs can be triggered by pushes, by a schedule or manually. Native integrations are provided for GitHub Actions, GitLab CI and other CI tools, and a CI integration can stop a pull request from merging when tests fail. Failures raise alerts through Slack, email or PagerDuty with logs and screenshots, and teams can download reproducible test scripts, logs, screenshots and videos. The vendor lists first class support for coding agents such as Claude Code, Cursor and Codex. Lark states that it is backed by Y Combinator and built by former Stripe engineers. Accounts can be created on the site or a demo requested. Pricing is not published.