LLM Fuzz CI

LLM Fuzz CI is a continuous integration tool implemented as a GitHub Action. It allows developers to mark tests to be 'fuzzed' by language model agents which generate adversarial inputs to try to break those tests. Inputs that cause test failures are replayed in a clean environment to ensure no false positives. The tool integrates with Python pytest and JavaScript vitest testing frameworks. It enables targeted AI-driven fuzz testing that understands code semantics to find complex bugs and vulnerabilities without noise from false alarms.