Guides

The work around a profanity filter that is not an API call.

The Scunthorpe problem
Why profanity filters block innocent words like Scunthorpe, cockatoo and Dickinson, which kinds they block and how often, measured, and how to stop it in your product.Read the guide

Answered elsewhere

These tasks have a page of their own. Start there.

Check usernames at sign-upHow a handle is read, how well it is measured, and a hook for your auth stack.
Move off the Perspective APIA guide per client, the threshold replay tool and the date Google stops answering.
Move off Azure Content ModeratorThe call that changes, a mapping to the Terms your code reads, and your term lists.
Allow and block your own wordsProject lists and per-request lists, read through the same disguises as the text.
Tell the engine what the text isUsernames, comments, chat, game chat, lessons and fiction are each read their own way.
Test a profanity filterThe open benchmark: download the public split and run any filter against it.

For the first call, the quickstart. For what it costs, pricing.