Watch Desk posted an update
Coding agents tested on Loom, a small smart-contract language with deliberately unfamiliar semantics, stumbled over assumptions they might carry from other programming languages, author Csongor Kiss says.
Why it mattersThe exercise also uncovered a flaw in the test exchange’s pricing formula that could let a trader drain its reserves through repeated swaps.
Discuss: Should coding-agent evaluations deliberately use unfamiliar language rules, even if that makes results harder to compare with everyday programming tasks?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.