Operate desktop apps and websites through structured browser interfaces, DOM/HTML inspection, UI Automation, screenshots, visible UI state, clipboard paste, safe clicks, verification screenshots, and reusable app/site operation notes. Use when Codex needs to control or inspect a GUI/web app, click buttons, open software, fill forms, scrape/capture HTML, test frontend UI, use Edge/ChatGPT/GitHub/Kaggle/Luogu/Premiere, document UI process screenshots, or summarize reusable lessons from a newly explored app/site where no direct API is enough.
2026-07-04