How to allow user-agent access to private discourse?

I don’t think we’ve ever encountered this situation before. You would need to authenticate the crawler as a regular user, probably through the Discourse API. Do we have any examples of this @techapj?