Playwright
Playwright in Practice
The automation framework that tests this site — what it checks, how it’s structured, and where it’s headed next.
Current Smoke Scan Status
The five checks that run against pinkitechi.com on every scan.
- Page load — every page returns and renders
- Console errors — no JS errors thrown on load
- Key elements — expected structural elements present
- Missing images — no broken image links
- Mobile render — layout holds at mobile widths
- Search — disabled until the search widget ships
How the Smoke Scan Works
A simplified look at what runs under the hood.
# playwright/smoke_scan.py def scan_page(page, url): page.goto(url) # check for console errors errors = collect_console_errors(page) # confirm key elements exist check_key_elements(page, url, config) # catch missing images check_images(page) return build_result(url, errors)
This runs once per page in qa_config.json‘s page list, with the WordPress.com tracking pixel excluded from image checks to avoid false failures.
Where This Is Headed: Multi-Tier Agents
The current smoke scan is one script. The planned structure splits testing into two tiers.
SA-01 → SA-07
Suite agents — broader test suites covering full user flows across the site, run less frequently.
PA-01 → PA-11
Probe agents — narrow, fast checks targeting specific elements or pages, run on every change.
findings-reporter.ts
Takes raw Playwright output and generates a findings summary, defects log, and workbook-ready results tab.
Common Gotchas
Things that have actually broken a scan here before.
New WordPress.com sites default to private, which blocks Playwright the same way it blocks search engines. Fix under Settings → Reading.
The built-in stats pixel can register as a “broken image” if it’s not explicitly excluded from image checks.
Switching permalink settings changes every page URL — any hardcoded paths in test config need updating to match.