Agent skill
webapp-testing
Toolkit for testing local web applications using Playwright. Use when verifying frontend functionality, debugging UI behavior, capturing browser screenshots, or viewing browser logs.
Install this agent skill to your Project
npx add-skill https://github.com/j0KZ/mcp-agents/tree/main/docs/universal-skills/webapp-testing
SKILL.md
Web Application Testing
This skill provides guidance for testing web applications using Playwright for browser automation.
When to Use This Skill
- Verifying frontend functionality works correctly
- Debugging UI behavior and interactions
- Capturing screenshots for documentation or debugging
- Testing form submissions and user flows
- Checking responsive design across viewports
- Validating API integrations from the frontend
What This Skill Does
- Browser Automation: Controls headless Chrome/Chromium for testing
- Screenshot Capture: Takes full-page or element screenshots
- Form Testing: Fills forms and validates submissions
- Network Inspection: Monitors API calls and responses
- Console Logging: Captures browser console output
- Responsive Testing: Tests across different viewport sizes
Decision Tree: Choosing Your Approach
User task → Is it static HTML?
├─ Yes → Read HTML file directly to identify selectors
│ ├─ Success → Write Playwright script using selectors
│ └─ Fails/Incomplete → Treat as dynamic (below)
│
└─ No (dynamic webapp) → Is the server already running?
├─ No → Start the server first, then run Playwright
│
└─ Yes → Reconnaissance-then-action:
1. Navigate and wait for networkidle
2. Take screenshot or inspect DOM
3. Identify selectors from rendered state
4. Execute actions with discovered selectors
Basic Playwright Script
from playwright.sync_api import sync_playwright
with sync_playwright() as p:
# Always launch chromium in headless mode
browser = p.chromium.launch(headless=True)
page = browser.new_page()
# Navigate to the application
page.goto('http://localhost:3000')
# CRITICAL: Wait for JavaScript to execute
page.wait_for_load_state('networkidle')
# Take a screenshot
page.screenshot(path='/tmp/screenshot.png', full_page=True)
# Your automation logic here
# ...
browser.close()
Reconnaissance-Then-Action Pattern
Step 1: Inspect the Rendered DOM
# Take a screenshot first
page.screenshot(path='/tmp/inspect.png', full_page=True)
# Get the full page content
content = page.content()
# Find all buttons
buttons = page.locator('button').all()
for btn in buttons:
print(f"Button: {btn.text_content()}")
# Find all links
links = page.locator('a').all()
for link in links:
print(f"Link: {link.get_attribute('href')}")
Step 2: Identify Selectors
From inspection results, identify the best selectors:
- Prefer
data-testidattributes if available - Use
text=for button text matching - Use
role=for ARIA roles - Fall back to CSS selectors
Step 3: Execute Actions
# Click a button by text
page.click('text=Submit')
# Fill a form field
page.fill('input[name="email"]', 'user@example.com')
# Select from dropdown
page.select_option('select#country', 'US')
# Wait for element to appear
page.wait_for_selector('.success-message')
Common Tasks
Form Testing
# Fill and submit a login form
page.fill('#email', 'test@example.com')
page.fill('#password', 'password123')
page.click('button[type="submit"]')
# Wait for navigation or response
page.wait_for_url('**/dashboard')
Screenshot Comparison
# Full page screenshot
page.screenshot(path='full-page.png', full_page=True)
# Element screenshot
page.locator('.hero-section').screenshot(path='hero.png')
# Specific viewport
page.set_viewport_size({'width': 375, 'height': 667}) # iPhone SE
page.screenshot(path='mobile.png')
Console Log Capture
# Capture console messages
def handle_console(msg):
print(f"Console {msg.type}: {msg.text}")
page.on('console', handle_console)
# Also capture errors
def handle_error(error):
print(f"Page error: {error}")
page.on('pageerror', handle_error)
Network Request Monitoring
# Monitor API calls
def handle_request(request):
if '/api/' in request.url:
print(f"API call: {request.method} {request.url}")
page.on('request', handle_request)
# Monitor responses
def handle_response(response):
if '/api/' in response.url:
print(f"Response: {response.status} {response.url}")
page.on('response', handle_response)
Best Practices
Wait Strategies
# Wait for network to be idle (best for SPAs)
page.wait_for_load_state('networkidle')
# Wait for specific element
page.wait_for_selector('.content-loaded')
# Wait for specific text
page.wait_for_selector('text=Welcome back')
# Timeout for slow operations
page.wait_for_selector('.data-table', timeout=30000)
Selector Best Practices
# Preferred: data-testid (explicit for testing)
page.click('[data-testid="submit-button"]')
# Good: Role-based selectors (accessibility)
page.click('role=button[name="Submit"]')
# Good: Text-based (user-centric)
page.click('text=Submit')
# OK: CSS selectors (when others unavailable)
page.click('.btn-primary')
# Avoid: Fragile selectors
# page.click('div:nth-child(3) > span') # Will break easily
Error Handling
try:
page.click('button.submit', timeout=5000)
except TimeoutError:
# Take screenshot for debugging
page.screenshot(path='/tmp/error-state.png')
print("Submit button not found - see error-state.png")
Common Pitfalls
| Pitfall | Solution |
|---|---|
| DOM not ready | Wait for networkidle before inspection |
| Element not found | Check if it's in an iframe or shadow DOM |
| Flaky tests | Add explicit waits, avoid arbitrary time.sleep() |
| Stale selectors | Use stable attributes like data-testid |
| Popup blockers | Configure browser context to allow popups |
WITH MCP Tools
If you have MCP tools available:
"Test the login flow on localhost:3000 and capture screenshots"
The webapp-testing MCP tool will handle server management and Playwright execution.
WITHOUT MCP Tools
Run Playwright directly:
# Install Playwright
pip install playwright
playwright install chromium
# Run your test script
python test_webapp.py
Installation
# Python
pip install playwright
playwright install chromium
# Node.js
npm install playwright
npx playwright install chromium
Tips
- Always use headless mode (
headless=True) for automation - Take screenshots at key points for debugging
- Use
networkidlewait state for dynamic apps - Capture console logs to catch JavaScript errors
- Test on multiple viewport sizes for responsive design
- Close the browser when done to free resources
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
debug-detective
Systematic debugging approach for ANY codebase, ANY language, ANY bug type
tool-discovery
Guide for discovering and using @j0kz MCP tools efficiently through search, filtering, and progressive exploration. Use when exploring available capabilities, finding the right tool for a task, or ...
project-standardization
Guides correct usage of @j0kz/mcp-agents standardization and automation scripts including version.json single source of truth, test count automation, URL casing rules, and critical workflow pattern...
api-integration
Master third-party API integration in ANY language with best practices and patterns
documentation-generation
Documentation generation and maintenance patterns including README structure, CHANGELOG management, API documentation, wiki synchronization, and badge updates. Use when creating package docs, updat...
competitive-ads-extractor
Extracts and analyzes competitor ads from ad libraries (Facebook, LinkedIn, TikTok, Google). Use when researching competitor messaging, creative patterns, campaign strategies, or ad inspiration. Ch...
Didn't find tool you were looking for?