All products

Test a local web app with AI

Check a real interaction in a local browser and save what appeared on screen. A server helper can start the project before the test runs.

To test a local web app with AI, describe an action and the visible result that should follow. With webapp-testing the agent can use a browser to exercise the actual interface. A checkout prototype might show a confirmation after a button press, for instance. A screenshot taken at that moment gives the designer something specific to inspect.

A static HTML file can be opened directly. For an application that renders content after loading, the agent can wait for the page to settle, inspect its elements and then interact. The included helper can manage more than one local server for a test that needs a frontend and backend.

What it does

  • Check whether a button exposes the expected panel.
  • Fill a test form and capture the resulting screen.
  • List visible links or fields once a dynamic page has loaded.
  • Save browser console messages while reproducing an issue.
  • Start and stop trusted local project servers around a test run.

Install and use

  1. Install the skill, Python Playwright and its Chromium browser in your development environment. Supply the project’s trusted start commands if needed.
  2. Ask “On the local booking form, enter dummy details, submit once and capture the confirmation.”
  3. Review the screenshot and any console messages against the intended behaviour.

FAQ

Source

Made by Anthropic. Original skill: webapp-testing, published on GitHub under the Apache-2.0 licence.

View on GitHub