Xe Iaso c460169047 feat(web): waste headless chrome bandwidth ...

Most of the worst of the worst scrapers run Headless Chrome. Headless
Chrome is difficult for Anubis to combat because it follows all the
rules that browsers do. The worst of the worst scrapers also use
residential proxy services. Those residental proxy services charge
upwards of $1 per GB of data egressed or ingressed. The Prompt API makes
Chrome download a 4Gi or 16Gi machine learning model. When you ask it to
start downloading, it will _continue_ downloading even when you leave
the Anubis challenge page.

This will make the local model answer "why is the sky blue?" in an
absurt amount of detail, which wastes both bandwidth and scraper CPU
(some scraping companies charge via Chrome CPU too).

Signed-off-by: Xe Iaso <me@xeiaso.net>

2026-05-06 17:27:16 -04:00

..

algorithms

chore: set up commitlint, husky, and prettier (#1451)

2026-02-15 08:19:12 -05:00

worker

chore: set up commitlint, husky, and prettier (#1451)

2026-02-15 08:19:12 -05:00

bench.ts

chore: set up commitlint, husky, and prettier (#1451)

2026-02-15 08:19:12 -05:00

main.ts

feat(web): waste headless chrome bandwidth

2026-05-06 17:27:16 -04:00