I’ve used Claude Code to add ordinary example-based tests, but I’m curious whether it can do a good job with property-based testing. For a parser or date-handling function, I’d rather check useful invariants across many generated inputs than end up with a handful of cases copied from the implementation.

Has it produced tests that actually uncovered bugs in your project? I’d like to hear what context or constraints helped, and whether you had to revise the generated properties substantially.