Lark is an end-to-end testing service for engineering teams. Tests are written in plain English and run on every commit, pull request or deployment, so a team learns within minutes if something broke. UI, API, CLI and mobile tests run on a continuous cadence. Lark was built by ex-Stripe engineers and is backed by Y Combinator.
Lark's agents can test anything a human could, including UIs, APIs, SDKs and async workflows. A user describes what to test and can choose deterministic runs or fully AI-driven flows. When a product changes, such as a dashboard revamp or a new API, tests adapt rather than breaking on selectors, and Lark alerts the team to the change.
When a test fails, Lark sends alerts through Slack, email or PagerDuty with detailed logs and screenshots, and users can download reproducible test scripts, logs, screenshots and videos. With a CI integration, failing tests can block pull requests from merging. Lark describes native integrations with GitHub Actions, GitLab CI and other CI tools, and support for coding agents such as Claude Code, Cursor and Codex through MCP.
Lark contrasts itself with Playwright, saying Playwright tests are code that is hard to write and maintain and covers the front end only, while Lark tests are written in natural language and can cover backend APIs. Users can create an account or request a demo.