Skip to content
SkillMixed833 tokens in use

webapp-testing

Anthropic's skill for testing local web apps: Claude writes native Python Playwright scripts, starts the dev servers with a helper, then inspects screenshots, the DOM and console logs.

197,764 stars on the repositoryanthropics/skills · skills/webapp-testing
/plugin install example-skills@anthropic-agent-skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing webapp-testing costs

35tokens idle
833tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Mixed

Script-ledMixedModel-led

Scripts do part of the work and the model's judgement does the rest.

This describes the skill's structure, how much of it is scripts. It does not say how well it works. Scripts: examples/console_logging.py, examples/element_discovery.py, examples/static_html_automation.py, scripts/with_server.py.

Touches

shelldetectedwrites filesdetected

Declared tags come from the skill's frontmatter; detected ones from its scripts and code blocks. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Excellent
  • Frontmatter11/15
  • Structure19/25
  • Examples15/15
  • Output10/15
  • Size15/15
  • Resources15/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent
  • Runs shell commands−5

    does this on the machine, and the frontmatter does not say so

    SKILL.md, line 41
  • Writes files−8

    does this on the machine, and the frontmatter does not say so

    examples/console_logging.py, line 31

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity72
  • Maintenance32
  • Community68
  • Maturity36
  • Momentum40

webapp-testing is Anthropic’s agent skill for interacting with and testing local web applications using Playwright. It has Claude write native Python Playwright scripts to verify frontend functionality, debug UI behaviour, capture browser screenshots and read browser console logs, and it ships one helper script, with_server.py, that starts one or more servers, waits for their ports to open, runs the automation and shuts the servers down again.

The SKILL.md gives Claude a decision tree for choosing an approach: read a static HTML file directly to find selectors, use the helper when the app’s server is not running yet, and follow a reconnaissance-then-action pattern when it is: navigate, wait for networkidle, take a screenshot or inspect the DOM, identify selectors from the rendered state, then act. It tells Claude to run the helper with —help and use it as a black box rather than reading its source into context, and points to three example scripts for element discovery, static HTML automation and console log capture.

What You Get

  • A decision tree for choosing between reading static HTML directly, starting servers with the helper, and inspecting an app that is already running
  • scripts/with_server.py, a helper that starts one or more server commands, polls each port until it accepts connections, runs your automation command and then terminates the servers
  • A template for an automation script using sync_playwright, headless Chromium and page.wait_for_load_state(‘networkidle’)
  • A reconnaissance-then-action pattern: full-page screenshot, page content and locator listing before selectors are chosen
  • Three example scripts: discovering buttons, links and inputs on a page, automating a local HTML file through a file:// URL, and capturing console messages to a log file
  • Best practices on descriptive selectors (text=, role=, CSS, IDs), explicit waits and closing the browser

Common Use Cases

  • Checking that a feature in a local dev server renders and responds to clicks and form input after a change
  • Debugging a UI bug by taking full-page screenshots and reading the browser’s console output
  • Testing a full stack locally by starting a backend and a frontend server together before the browser script runs
  • Automating a static HTML page or prototype opened from disk through a file:// URL
  • Listing the buttons, links and input fields on a page to find reliable selectors for a test

What Makes It Unique

The other Anthropic skills listed here mostly shape what Claude produces, like frontend-design for interfaces or pdf and docx for documents, or guide building something, like mcp-builder for MCP servers. webapp-testing is about checking a web app that already exists, by driving a real browser against it. It is small and practical: one server helper treated as a black box so its source stays out of the context window, a single rule it calls critical (wait for networkidle before inspecting a dynamic app), and a look-first pattern where selectors come from the rendered page rather than from guessing at the source.

Under The Hood

SKILL.md is a short guide that Claude follows when writing its own Python Playwright scripts, which launch Chromium in headless mode, open localhost URLs or file:// paths, click, fill forms, take screenshots and read page content. The bundled scripts/with_server.py takes one or more —server commands with matching —port values, starts each command through a shell (so values like “cd backend && python server.py” work, and whatever command is passed is run as given), polls localhost on each port until it accepts a connection or a timeout passes, then runs the command after — and terminates or kills the servers when it finishes. The skill does not install anything itself; it assumes Python and the Playwright package with a Chromium browser are already present. The example scripts write screenshots and a console log to files, under /tmp and /mnt/user-data/outputs, and the browser only reaches the local URLs the scripts point it at.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers