Post

Apparently getting Claude to write \u-escaped versions of Unicode code points in Kotlin string literals (as opposed to the literal Unicode characters) is very difficult. It'll keep retrying with literal code points, and when it doesn't do that, it'll use extra backslashes