Routes browser automation intelligently by classifying tasks as static-capable versus browser-required before choosing tools. Maintains domain-specific notes in `references/site-patterns/` that capture login quirks, selector patterns, and anti-bot defenses for future runs. Loads targeted reference files based on task type rather than guessing - debug handoffs for DevTools contexts, control plane workflows for CMS tasks, deep DOM guides for Shadow DOM work. The routing logic prevents silent downgrades from browser automation to basic HTTP fetches when live session state actually matters. Honestly, the prerequisite checks and decision trees make this more systematic than most browser automation, though the reference file system adds complexity for simple tasks.
npx skills add xixu-me/skills --skill use-my-browser