Webapp Testing
Developer Tools
B rating
License: Apache-2.0 (skill LICENSE.txt, rev dbd4588)
An official Anthropic web-testing skill that guides an agent to use Python Playwright for local page checks, element discovery, screenshots and browser logs. Its with_server.py helper can start development servers, wait for ports, run tests and stop the processes it starts. It has real command and browser-action capabilities, so use an isolated test project rather than production accounts. This site performed only a static review of the pinned commit, not installation, server execution or end-to-end tests.
Claude Code
Usage instructions
The download is the complete original repository ZIP at commit dbd4588f9e1033efb41dad4bef2f7947c8993d44. Take only skills/webapp-testing and retain its LICENSE.txt, scripts and examples; other directories may have different licenses. The skill declares no separate semantic version, so this listing identifies the commit.
- Place it in
.claude/skills/webapp-testing/ in a Claude Code project. Read SKILL.md, with_server.py and the examples first. Verify file, terminal and browser permissions separately for other hosts; SKILL.md alone does not prove compatibility. - In an isolated Python virtual environment, install
python -m pip install playwright, then run python -m playwright install chromium. These commands download software; verify sources and local policy first. Downloading the skill does not perform them automatically. - If a local development server already exists, allow tests only on approved local addresses. If using the helper to start one, first run
python scripts/with_server.py --help. Supply only a reviewed project startup command, the port and the test script as documented. - Example targets and paths are placeholders. Change URLs, buttons and output directories first. On Windows in particular, replace /tmp and /mnt/user-data paths and create an appropriate output directory.
- Start with: “Check required-field messages on the local test form and save screenshots and console errors. Do not log into real accounts, send messages or make payments.” Use test data and low-privilege accounts, and review the actual script before execution.
Safety review
Rating B is this site's static editorial assessment of a pinned commit, not a sandbox guarantee or complete security certification. SKILL.md, with_server.py and three examples were reviewed. The helper executes the supplied server command with shell=True, runs a test command and terminates or kills the processes it starts. Examples read pages, click controls, fill forms and write screenshots or logs.
- Do not concatenate untrusted page text, model output or user input into --server commands. Startup commands may install dependencies, access the network or change files; none were executed during this review.
- Use an isolated test project and dedicated ports, not production services. Helper shutdown does not guarantee every descendant process has exited. Check logs and process state for long-running tasks.
- Browser actions may submit real data. Screenshots, logs and login state may contain sensitive information and should not be uploaded or published. A person must confirm allowed targets and output directories.
- Waiting for networkidle alone does not establish readiness for every application, especially persistent-connection pages. Wait on actual element state and task-specific acceptance criteria.
Copyright and takedown notice: AI Islands curates this page from public information. Skills, code, documents and packages remain the property of their original authors or rights holders. This listing is provided for indexing, research and installation convenience. If you believe any listing or download link infringes your rights, contact ai-islands@streamflowintel.com with proof of ownership, relevant URLs and your request. We will review and remove or adjust the content promptly. Review package permissions, dependencies and safety risks before installing.