Playwright

Playwright now exposes page.webmcp and frame.webmcp to list and call tools a page registers via the experimental WebMCP browser API, and Playwright MCP and playwright-cli support calling these tools directly for agent testing; it also adds video fps and CSS-styled action decorations, locator.within() for combining locators, and new APIs like page.getByRef(), apiRequestContext cookie methods, and a lock option for test.describe.configure(). This is a breaking release because the screen property is now forwarded from device descriptors, so window.screen and media queries reflect emulated screen size by default, requiring explicit screen: undefined in config to opt out, and JSX in test files now follows tsconfig.json settings instead of previous defaults.

Matters for: Teams building or testing pages that expose tools via the experimental WebMCP browser API, and teams relying on device descriptor emulation (like devices['Desktop Chrome']) for screen-size-sensitive tests.

Playwright 1.63 lets tests declare named locks so tests sharing a resource never run concurrently while everything else still runs in parallel, adds frameLocator() calls that search across all frames without first locating the iframe, introduces locator.visible() to match only visible elements, and lets tracing capture aria and screen snapshots alongside DOM snapshots for side-by-side inspection in the trace viewer. Test steps can now carry structured subtitle and params data for reporters, and Playwright API steps automatically report the target locator or navigation URL.

Matters for: Useful for teams running large parallel end-to-end suites against shared external services or accounts, and for teams testing pages with nested iframes or needing detailed accessibility-focused trace debugging.