⚡ EXECUTION INSTRUCTIONS FOR CLAUDE

When this command is invoked, YOU (Claude) must execute these steps immediately: This is NOT documentation - these are COMMANDS to execute right now. Use TodoWrite to track progress through multi-phase workflows.

🚨 EXECUTION WORKFLOW

Phase 1: Manual test execution

Action Steps: With Puppeteer mode active:

1. Server runs on http://localhost:6006?test_mode=true&test_user_id=test-user-123
2. Use MCP functions for navigation, clicking, filling forms
3. Capture screenshots for validation
4. Execute JavaScript for complex interactions

📋 REFERENCE DOCUMENTATION

/puppeteer Command

Purpose: Set Puppeteer MCP as the preferred browser automation tool for Claude Code CLI sessions

Usage: /puppeteer

Action: Configure session to use Puppeteer MCP instead of Playwright for browser testing

What it does

1. Sets session preference to use Puppeteer MCP for all browser automation
2. Adds --puppeteer flag to test runner commands automatically
3. Updates test execution to use MCP tools instead of Playwright
4. Enables screenshot capture through Puppeteer MCP functions

Benefits

• ✅ No dependency issues - Uses Claude Code's built-in MCP
• ✅ Real browser automation - Not HTTP simulation
• ✅ Direct integration - Native to Claude Code environment
• ✅ Visual validation - Built-in screenshot capabilities
• ✅ JavaScript execution - Full browser scripting support

Integration with test runners

When /puppeteer is active, test commands will automatically:

• Use ./run_ui_tests.sh mock --puppeteer instead of Playwright
• Execute browser tests via MCP functions
• Capture visual evidence through screenshots
• Provide test server with proper auth bypass

Usage in session

# Set Puppeteer preference

/puppeteer

# Now all browser tests use Puppeteer MCP

/testui  # Uses --puppeteer flag automatically

Example workflow

# Navigate to test application

mcp__puppeteer-server__puppeteer_navigate(url="http://localhost:6006?test_mode=true")

# Take screenshot

mcp__puppeteer-server__puppeteer_screenshot(name="initial_state")

# Fill form field

mcp__puppeteer-server__puppeteer_fill(selector="#campaign-title", value="Test Campaign")

# Click button

mcp__puppeteer-server__puppeteer_evaluate(script="document.querySelector('#next-btn').click()")

# Validate result

mcp__puppeteer-server__puppeteer_screenshot(name="after_action")

Compatibility

• Works with all existing test scenarios
• Replaces Playwright for browser automation
• Maintains same test URLs and auth bypass
• Compatible with structured fields testing