Goodreads returns an empty HTML page to headless browsers it identifies
as automated (navigator.webdriver=true, AutomationControlled feature).
This caused the Export Library button to never be found.
- Launch Chromium with --disable-blink-features=AutomationControlled
- Set a realistic user agent and viewport on the browser context
- Strip navigator.webdriver via an init script
- Guard the logged_in check so a blank page isn't mistaken for a session
- Wait for networkidle before querying the export button, and use
wait_for(visible) instead of is_visible() to tolerate JS render delay
- Add CSS class fallback (button.js-LibraryExport) if role lookup fails
- Add AGENTS.md documenting the bot detection pattern and export flow
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Replace the old fixed 10-minute wait with readiness-based waits on the export link state, so exports start downloading as soon as Goodreads marks them ready.
Also harden login selectors/timeouts, add fallback CSV fetching when browser download events do not fire, and add CLI modes for --check-login, --login-only, --no-headless, and --har-path.
Closes#3
