Inside the Microsoft Playwright Plugin for Claude Code

What is the microsoft/playwright plugin, actually

Playwright is Microsoft's own framework for web testing and automation, built to run one API against Chromium, Firefox, and WebKit. microsoft/playwright is the Claude Code plugin published straight from that project's own GitHub repository (95,114 stars, by Flowy's own count), not a wrapper someone else built around it. Flowy indexes the repository the way it indexes every crawled plugin: it reads the manifest and lists what actually ships inside; it does not own, bundle, or host the code. Flowy files this listing under the Testing category.

The plugin ships 14 components in total: 7 agents and 7 skills. Read together, they split into three groups: prompts and agents that plan, generate, and fix tests for you; skills for everyday Playwright work such as the CLI and component testing; and skills built for people who work on the Playwright project itself. Install it from the listing page linked at the top of this guide, since that page generates the current install command rather than a copy that can drift out of date.

Which agents plan, generate, and fix tests

Six of these agents form three matched pairs: a prompt file and an agent file that do the same job. playwright-test-plan.prompt creates a test plan, and playwright-test-planner.agent is described as the one to use when you need a complete test plan for a web application or website. Use either when you are starting from a page or a feature and need to decide what to test before writing an assertion.

playwright-test-generate.prompt and playwright-test-generator.agent sit on the generation side. The prompt generates a test plan; the agent is described as the one to use when you need to create automated browser tests using Playwright.

playwright-test-heal.prompt and playwright-test-healer.agent fix tests. The agent is described as the one to use when you need to debug and fix failing Playwright tests, the failure mode that shows up whenever a selector or a flow changes.

The seventh, playwright-test-coverage.prompt, has no matching agent file. It produces test coverage, for a reader who wants to know how much of an application a Playwright suite already reaches.

Which skills cover everyday Playwright work

playwright-cli is the general-purpose one: it automates browser interactions, tests web pages, and works with existing Playwright tests. It is the skill most people reach for first.

playwright-component-testing is scoped to a single component rather than a whole page. It sets up component testing with Playwright using a story gallery, scaffolding stories and a gallery dev page driven by Playwright's built-in mount fixture instead of a separate component-testing runtime. Its own description points to testing React or Vue components in isolation, and to moving off another component-testing setup.

Which skills help once something has already gone wrong

playwright-trace inspects Playwright trace files from the command line: actions, requests, console output, errors, snapshots, and screenshots.

playwright-test-results queries Playwright CI test results from an aggregated DuckDB database, answering questions about flaky tests, failure rates, and slow tests, or the results for one run, one commit, or one pull request, without hunting through GitHub artifacts by hand.

playwright-triage reproduces a bug report from the information already in a GitHub issue, whether the issue is new or just picked up another comment.

Which skills are for people building Playwright itself

Two skills are aimed at Playwright's own maintainers rather than at a team testing its own application. playwright-devops covers CI failure analysis, workflow debugging, and release operations for the project. playwright-dev explains how to develop Playwright itself: how to add APIs, MCP tools, CLI commands, and vendor dependencies. Neither does much for you unless you are working inside the Playwright codebase.

Is this plugin for testing an app, or for working on Playwright itself

Both, and the plugin draws a clean line between the two. Most of it, the test planning, generating, and healing agents, plus playwright-cli, playwright-component-testing, playwright-trace, playwright-test-results, and playwright-triage, is for any team that already tests with Playwright and wants Claude Code inside that loop. playwright-devops and playwright-dev are for contributors to the Playwright project itself. If you only test your own application, the first group is the reason to install this plugin, and the other two skills will sit unused.

How do I install it

Flowy indexes microsoft/playwright from its public GitHub repository; it does not host or bundle these components. Install from the listing page linked at the top of this guide rather than from a command written into this text, since the listing page is what stays current. Once installed, match the group above to what you are actually trying to do: most readers only need the everyday testing skills and the test generation and healing agents, not the maintainer tooling.