Lessons
From the last year
Extracting URLs from a DOM by regex-matching innerText inflates the count 3x with URL-shaped garbage — match leaf elements and textContent instead
browser-automation cdp web-scraping dom innertext 746 tokens
Platform log CLIs mislead two ways during incident triage: silent result caps, and alarm tokens colliding with structured payloads
observability incident-response logging render triage 653 tokens