Playwright best practices skill: a testing reference for your coding agent

Source: currents-dev/playwright-best-practices-skill · Maintainer: community: Currents.dev · License: MIT · Checked: 23 October 2026

What is the Playwright best practices skill?

The Playwright best practices skill gives a coding agent a large, organized reference for writing, debugging and maintaining Playwright tests in TypeScript, so the tests it writes use stable locators, proper waits and sound structure instead of guesswork. It is for teams that already use Playwright Test, or are starting to, and want an agent like Claude Code, Codex or Cursor to follow good testing habits.

It comes from Currents.dev, which makes a dashboard for Playwright test results, and lives in currents-dev/playwright-best-practices-skill. The skill's description is a long list of triggers. It begins "Use when writing Playwright tests, fixing flaky tests, debugging failures, implementing Page Object Model, configuring CI/CD" and continues through authentication, accessibility, file uploads, time mocking, WebSockets, mobile layouts, GraphQL, security testing, Electron apps and browser extensions, ending: "Covers E2E, component, API, visual, accessibility, security, Electron, and extension testing."

As of 23 October 2026 the repo had about 400 GitHub stars and was last pushed on 21 July 2026, but its skills.sh page listed about 97,700 installs, so far more people install it than star it.

What does the Playwright best practices skill do?

It routes, then reads. The SKILL.md itself holds little test advice. It is a map: tables grouped by activity (writing new tests, mobile, browser APIs, debugging, error cases, multi-user, architecture, frameworks, refactoring, infrastructure) that name which reference file to open, followed by a decision tree. The README explains the idea: "the AI is directed to the right reference depending on what you're doing, so you get focused advice without loading everything at once." That matters, because the references add up to roughly 80,000 words.

It ends with a validation loop. After writing or changing tests, the agent should run npx playwright test --reporter=list, read failures and traces (npx playwright show-trace), fix and re-run, only move on when everything passes, and run critical tests several times with --repeat-each=5.

The advice is concrete. For example, core/locators.md sets a priority order: role-based getByRole first, then label and placeholder, then text, then test IDs, with CSS and XPath as a last resort. debugging/flaky-tests.md sorts flakiness by type and walks through reproducing it, finding the root cause and fixing it.

What's in the folder: one SKILL.md plus 58 Markdown references in eight folders (the README says 57):

• core/ (11): test structure, locators, assertions and waiting, Page Object Model, fixtures and hooks, test data, annotations, configuration, global setup, projects, test tags.
• testing-patterns/ (15): accessibility with axe-core, API, component, visual regression, forms, drag and drop, file operations, file uploads and downloads, GraphQL, canvas and WebGL, i18n, Electron, browser extensions, security, performance.
• advanced/ (8): authentication and complex auth flows, mobile, clock mocking, multi-tab and popups, multi-user, advanced networking, third-party services.
• infrastructure-ci-cd/ (9): CI basics, GitHub Actions, GitLab, other providers, Docker, sharding, performance, reporting, coverage.
• debugging/ (4), browser-apis/ (4), frameworks/ (4: React, Angular, Vue, Next.js) and architecture/ (3: POM vs fixtures, test types, when to mock).

There are no scripts.

How do you install the Playwright best practices skill?

With the skills CLI. The README's only install command uses the skills CLI:

npx skills add https://github.com/currents-dev/playwright-best-practices-skill

The repo holds one skill, so there is nothing to pick. Add -g to install for every project, or -a claude-code to skip the agent prompt. For Claude Code the folder lands in .claude/skills/playwright-best-practices/, or ~/.claude/skills/playwright-best-practices/ with -g, the standard Claude Code skill locations. Install it per project if only some of your repos use Playwright.

Other agents. Codex and Cursor read the same SKILL.md format; see Codex skills and Cursor skills. The repo has no Claude Code plugin manifest.

claude.ai. Zip the playwright-best-practices folder and upload it under Customize, then Skills, as the Claude Help Center describes. Keep the subfolders, since the SKILL.md links to them by relative path.

How do you use it?

You don't need to name the skill. The README says to describe the task, "fix this flaky login test" or "add accessibility tests", and the agent will use it when relevant. Example prompts (suggestions, not quotes from the repo):

• This login test passes locally and fails in CI about one run in five. Find out why and fix it.
• Refactor tests/e2e to use fixtures instead of page objects where it makes sense, and explain the choice.
• Add Playwright tests for the password reset flow, including the email step, without hitting a real inbox.
• Set up sharded Playwright runs in GitHub Actions with an HTML report as an artifact.
• Write a test that freezes time to 31 December and checks the year-end banner.

Requirements. A project with Playwright Test, which Playwright's installation docs set up. The skill needs no API keys or extra packages; some references show optional libraries such as @axe-core/playwright for accessibility checks.

Tips and limits

From the files, not from testing:

• It is advice, not hands. The skill can't open a browser. Pair it with playwright-cli, whose test-generation workflow writes and heals tests, while this skill keeps them well structured.
• TypeScript only. The README targets Playwright in TypeScript. Python and Java teams will get the principles but not usable code.
• Check it against Playwright's docs. The guidance is Currents' own. Where it touches core rules, compare with Playwright's best practices page, which is the official word and updates with each release.
• Repeat runs cost time. The loop's --repeat-each=5 is meant for critical tests. Don't let the agent apply it to a whole suite on every change.
• Long trigger list. The description matches many testing phrases, so in a repo with other testing skills the agent may load this one more often than you expect. Install it only where you use Playwright.

Security audits. On 23 October 2026, its skills.sh page showed Pass from all three listed audits (Gen Agent Trust Hub, Socket and Snyk). The skill is Markdown only, but still read the files before installing. The MIT License lets you copy and edit it freely as long as you keep the notice.

Save it in your Kipt library

In Kipt, open Skills and paste the skill's GitHub folder link into the import box. Kipt reads the SKILL.md and its reference folders. You can also drop the folder or a .zip. Your skills live in your own private GitHub repo and install into Claude Code or Codex with one command:

npx skills add https://github.com/<you>/<repo>.git --all -g

Kipt is not affiliated with Currents.dev.

Similar skills

• Playwright CLI skill: Microsoft's official skill for driving the browser and generating tests.
• webapp-testing skill: Anthropic's lighter Python approach to checking a local app.
• TDD skill: Matt Pocock's red-green loop for writing the test first.
• Accessibility skill: fix WCAG issues, then lock them in with this skill's axe-core tests.