What is Browserbase?
Browserbase supplies remote browsers that customers can use for scraping, form submission, testing, and AI agent tasks. A Browserbase request is therefore part of a customer's browser session. It is not evidence of one centrally managed Browserbase index visiting every site for the same purpose.
That distinction changes how the traffic looks and how it should be interpreted. A session can navigate pages and perform actions that an ordinary HTML crawler would never attempt. The user agent tells you which browser platform carried the request, but it does not tell you whether the customer was testing a checkout, extracting a page, or asking an agent to complete a task.
Browserbase is infrastructure, not a documented training-data collector or AI search index. Allowing it does not by itself place a page in a model's training set, and blocking it does not remove a site from a named search product. A customer could still use material retrieved through the browser in an AI workflow, so the downstream use depends on that customer's task.
Cloudflare's bot directory lists Browserbase as an agent operated by Browserbase and links to https://browserbase.com. It also provides https://www.browserbase.com/.well-known/http-message-signatures-directory for Web Bot Auth. A valid HTTP message signature can establish that Browserbase carried the session, although it cannot explain the customer's intent.
The published robots signals are mixed. Browserbase's own data-retrieval guidance tells customers to check a site's robots.txt, while Cloudflare's bot directory marks the Browserbase agent as not following robots.txt. Treat a rule for the Browserbase token as a clear policy statement, not as guaranteed enforcement for every customer-run session.
Use the token, signature result, requested paths, and any attempted actions together when reviewing this traffic. The robots.txt guide explains the voluntary crawl rules; authenticated routes and edge controls remain the reliable boundary for browser automation.
