Shiplight AI is an agent-native end-to-end testing platform built around the idea that tests should be authored and maintained by the same coding agents developers already use. It installs into tools like Claude Code, Cursor, Codex, VS Code and Antigravity, letting an agent walk your application in a real browser and generate tests written as plain-English YAML that describes user intent rather than CSS selectors. Those tests live in your repository and can be reviewed in a pull request like any other code change.
The platform emphasizes self-healing. When a cached locator goes stale — a button moves or a class is renamed — Shiplight re-reads the page, resolves the original intent with AI, and updates its cache mid-run, so intended UI changes do not break the suite while genuine regressions still fail loudly. Settled steps replay deterministically from a cache at full Playwright speed, meaning only new or changed screens cost a model call, so mature suites run faster and cheaper over time.
Shiplight also covers execution and triage: the same test file runs locally, on Shiplight-hosted CI runners, or on your own infrastructure, and results land on the pull request with logs, screenshots and traces. A fix workflow can reproduce a failure, find the cause and update the test. It is positioned as enterprise-ready with SOC 2 compliance, audit logs, private cloud/VPC deployment options, and a free starting tier. Pricing for teams is available on request.
