Skip to content
HUIAHUB

anthropics/skills/skills/webapp-testing

Web app testing with Playwright

A toolkit for testing local web applications with Playwright: checking frontend behavior, debugging the UI, taking browser screenshots and reading browser logs.

Agent skillAgent skills
View the skill on GitHub (opens in a new tab)

What Web app testing with Playwright does

The skill has the agent write native Python Playwright scripts to test a web app on the local machine. It starts with a decision. For a static HTML file, the agent reads the file to find selectors and writes a script against them. For a dynamic app that is not running yet, it uses the bundled scripts/with_server.py, which starts one or more servers (a backend and a frontend, say), waits until their ports answer, runs the automation script and stops the servers afterwards. The SKILL.md tells the agent to run that helper with --help and treat it as a black box instead of reading its source.

Once the app is up, the skill follows a look-first pattern. The agent opens the page in headless Chromium, waits for network activity to settle, takes a screenshot or reads the rendered DOM, picks selectors from what it finds and only then clicks or types. Three example scripts show how to list the buttons, links and inputs on a page, automate local HTML through file:// URLs and capture console logs. What you get back is a Playwright script and its output: screenshots, page content and browser console messages.

Install

/plugin install example-skills@anthropic-agent-skills

Read the security check below before you run it. The repository's own instructions win if they differ.

Security check

What the skill's files do, from reading them. A flag that nobody has checked yet says "Not checked", never "No".

Runs scripts
Yes
Uses the network
Yes
Installs hooks
No
Asks for secrets
No

The skill ships four Python files: the scripts/with_server.py helper and three examples. The helper runs each server command it is given through the shell, so the agent can start anything it writes there, such as npm run dev. The scan flags network use because the instructions load URLs; the examples point at localhost and local files. The scan found no hooks and no secret handling. Before installing, read with_server.py, and check each server command your agent proposes before you approve it.

Scan details: Ships 4 script files: examples/console_logging.py, examples/element_discovery.py, examples/static_html_automation.py, scripts/with_server.py. SKILL.md instructions include web requests or URLs to fetch. Checked at commit 8a1541c4a3ff.

Use cases

  • Check that a local frontend change works in a real browser.
  • Capture full-page screenshots of a running dev server.
  • Collect browser console logs while reproducing a UI bug.
  • Test an app that needs a backend and frontend started together.

Limitations

  • Built for local web apps, using Python scripts with Playwright and Chromium.
  • The SKILL.md does not explain how to install Python or Playwright.
  • It warns that inspecting a dynamic page before network idle gives wrong results.
  • Servers that do not answer on their port within the timeout (30 seconds by default) stop the run.

Sources