The stories from this are pretty fun and silly (as long as it's not excessively horrifying to you how much copyrighted content is baked into the models) gist.github.com/vcarl/025a02...
Extremely anecdotal, not actual work, but I had four independent Sonnet agents playing an MMO with access to self-modify the SDK that hit the game's API. They used in-game chat to discuss bugs they noticed and notify each other of commits they pushed, and a bunch of them were real fixes