microsoft/playwright
3 skills2.7k installs272k starsGitHub
Install
npx skills add https://github.com/microsoft/playwrightSkills in this repo
1Playwright DevPlaywright Dev is an agent skill for intermediate-to-advanced contributors who hack on Microsoft's Playwright repository—not merely write end-user tests. It routes you through the monorepo map in CLAUDE.md and deep dives on library architecture, adding or changing APIs, registering MCP tools and CLI commands, managing vendored dependencies, updating WebKit Safari version strings in wkBrowser.ts, bisecting regressions across published npm versions, and working on the dashboard behind playwright cli show. Solo builders shipping agent products that fork or extend Playwright benefit when they need correct DEPS.list hygiene, protocol-layer changes, and test coverage alongside feature work. The skill is documentation-forward meta guidance: it does not run your app's E2E suite; it teaches how the framework is structured so changes land in the right packages with the right checks.2.1kinstalls2Playwright Mcp Devplaywright-mcp-dev is a contributor-oriented agent skill for developers extending Microsoft’s Playwright MCP server and terminal CLI—not for end users writing product E2E tests alone. It documents the exact package layout under packages/playwright: where browser tools live, how to register capabilities, how CLI categories map to MCP tools, and which spec files exercise each surface. Testing discipline is explicit—npm run ctest-mcp with a category argument, no test --debug—and doc updates must touch SKILL.md plus references when behavior changes. Solo builders forking Playwright or shipping a custom MCP distribution use this to avoid breaking the tool registry or shipping undocumented flags. It pairs naturally with broader Playwright testing skills for consumer test authoring, but this skill is meta: it tells your coding agent how to change the agent bridge itself.322installs3Playwright Devopsplaywright-devops is a workflow skill for maintainers and solo builders who own a Playwright-heavy repository and need a structured picture of CI health after a main-branch commit. Phase one runs the bundled fetch script to download failed job logs into a local commit folder with summary.json metadata and per-job .log files. Phase two reads those artifacts, extracts failing test names and errors, and writes a markdown CI Health Report that leads with a summary and groups issues by likely root cause—flakes, infrastructure, platform-specific problems, or possible commit regressions when evidence supports it. It is scoped in the SKILL.md to microsoft/playwright’s main branch workflow but the pattern applies to any repo that copies the script layout. Use it when CI is noisy and you need one narrative for the tree, not a single test rerun. Intermediate complexity: you must run bash, read JSON logs, and interpret long CI output.230installs