Configure and observe automatic CAPTCHA solving on a Browserbase session, including disabling it for human-in-the-loop flows

Documented steps

1. Authenticate with X-BB-API-Key on POST https://api.browserbase.com/v1/sessions; CAPTCHA behavior is configured under browserSettings. Reference: https://docs.browserbase.com/reference/api/create-a-session
2. Know the default: browserSettings.solveCaptchas is a boolean that defaults to true, so Browserbase attempts to solve CAPTCHAs automatically unless you opt out.
3. Set browserSettings.solveCaptchas: false to disable automatic solving - do this when you want a human to complete the challenge through an interactive live view, or when the target's terms prohibit automated solving.
4. For non-standard CAPTCHA widgets, set browserSettings.captchaImageSelector to a CSS selector for the challenge image element.
5. Set browserSettings.captchaInputSelector to a CSS selector for the answer input field so the solver knows where to type.
6. Detect solve activity at runtime by fetching GET https://api.browserbase.com/v1/sessions/{id}/logs with header X-BB-API-Key, or via the SDK bb.sessions.logs.list(sessionId).
7. Look for the console log strings browser-solving-started and browser-solving-completed to mark the beginning and end of an automatic solve. Reference: https://docs.browserbase.com/platform/browser/observability/observability
8. Build waits around those markers rather than fixed sleeps: a solve in progress can otherwise look like a hung page to your automation.
9. If solves fail repeatedly, address the upstream cause - pair solveCaptchas with browserSettings.verified: true (Scale plan) and residential proxies so fewer challenges are served in the first place.
10. For human-in-the-loop, combine solveCaptchas: false with an interactive live view iframe from /v1/sessions/{id}/debug so an operator can intervene.