TestSprite Open-Sources AI Code Verification CLI Tool

TestSprite has released an open-source command-line tool that lets AI coding agents verify their own work, addressing a growing problem where autonomous agents ship buggy or incomplete code. The tool runs tests in the cloud like a real user, returning detailed failure reports so agents can fix and rerun in a continuous quality loop. Alongside the release, TestSprite launched CoderCup, a public AI coding competition using the CLI as a neutral referee. Results showed smaller model Kimi achieved the highest correctness score at 0.89 despite being the slowest and cheapest, while every agent tested broke previously working features.