webapp-testinglisted

# Web Application Testing Test local web applications by writing native Python Playwright scripts, or by using the browser MCP tools available in VS Code Copilot Chat (open_browser_page, click_element, read_page, screenshot_page, type_in_page). This skill teaches the **reconnaissance-then-action** pattern and the **static vs dynamic decision tree**. It does not bundle any helper scripts — the agent writes Playwright code directly or uses the browser MCP tools. ## Decision Tree: Choosing Your Approach ```mermaid flowchart TD A[User task] --> B{Is it static HTML?} B -- Yes --> C[Read HTML file directly<br/>to identify selectors] C --> D{Success?} D -- Yes --> E[Write Playwright script<br/>using selectors] D -- Fails/Incomplete --> F[Treat as dynamic] B -- No, dynamic webapp --> G{Is server already running?} G -- No --> H[Tell user to start server<br/>or use MCP browser tools] H --> I[Reconnaissance-then-action] G -- Yes --> I I --> J[1. Navigate and wait for networkidle] J --> K[2. Take screenshot or inspect DOM] K --> L[3. Identify selectors from rendered state] L --> M[4. Execute actions with discovered selectors] ``` ## Two execution paths ### Path A: Native Playwright (Python) Write a self-contained Playwright script. The agent manages server lifecycle itself — it does NOT shell out to a bundled helper. If a dev server needs to start, the user starts it in a separate terminal, or the agent uses the run_in_terminal t