You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
The browser-based scraper is now fully blocked on GitHub Actions. As of ~2026-07-09 the Fastly bot-challenge fingerprints the automated browser itself — it blocks headless and headed, Playwright's Chromium and real Chrome, with navigator.webdriver=false + stealth scripts, even a persistent real-Chrome profile. Only a genuine human-driven browser passes, so this is not solvable on our side within Actions (see #27, #26, and thunderbird/github-action-thunderbird-aaq#34).
We have no path forward on GitHub Actions. We need to move off it.
Proposed path forward
Stand up a self-hosted runner with a static public egress IP — a VPS or Thunderbird infra.
Summary
The browser-based scraper is now fully blocked on GitHub Actions. As of ~2026-07-09 the Fastly bot-challenge fingerprints the automated browser itself — it blocks headless and headed, Playwright's Chromium and real Chrome, with
navigator.webdriver=false+ stealth scripts, even a persistent real-Chrome profile. Only a genuine human-driven browser passes, so this is not solvable on our side within Actions (see #27, #26, and thunderbird/github-action-thunderbird-aaq#34).We have no path forward on GitHub Actions. We need to move off it.
Proposed path forward
/api/2/is reachable directly over HTTP; no browser needed..github/workflows/scrape.ymlto run on the self-hosted runner.An API token that bypasses the challenge would be an acceptable alternative to the IP allowlist — either mechanism unblocks us.
Open questions / decisions
sumo.py(challenge-cookie logic can be removed once the IP is exempt).Context
cc @rtanglao @mzeier