Static analysis for AI-generated code.
Grounded in available product and source data
Hallint is static analysis for AI-generated code, aimed at the mistakes coding assistants repeat rather than the ones human-written code produces. The project's own documentation describes the failure modes it targets, and every rule carries a severity grade and a suggested fix.
It ships as a command line program and as a library, so it can run over a folder in a terminal or inside your own tooling. Rules can be switched on per project, suppressed inline where a pattern is deliberate, and declared public for routes that are meant to stay open.