Hey Product Hunt — I built rea-captcha after watching browser agents get most of the way through a task, hit a verification page, and lose the session. The awkward part was not detecting a CAPTCHA. It was deciding when the agent could safely continue. A click is not proof that the protected page accepted the verification, and a late success from an older attempt should not release a newer gate. rea-captcha keeps the existing Playwright/CDP page, pauses for either an authorized test action or a bounded human handoff, then checks fresh page evidence before resuming. The live demo uses official non-production keys for reCAPTCHA, Turnstile, and hCaptcha, plus one owned image/text case. Unknown production challenges stop for the user; this is not a generic bypass. I’d especially value feedback from people building with Browser Use: what should count as a safe resume signal in your workflow?

Forum Threads

p/rea-captcha

•

12h ago

When should a browser agent stop and ask for help?

While building rea-captcha, I found the hard part wasn t detecting a CAPTCHA. It was deciding what counts as enough evidence to let the original Browser Use task continue.

My current rule is simple:

keep the browser session the agent already owns

View all

rea-captcha

Safe CAPTCHA handoff for Browser Use agents

Safe CAPTCHA handoff for Browser Use agents

Forum Threads

When should a browser agent stop and ask for help?

Forum Threads

When should a browser agent stop and ask for help?