Catala (Merigoux et al. 2021) is a recent domain specific language that aims to simplify translation of legal statutes describing algorithmic decision processes (e.g., tax law or social benefits) into executable code. Its semantics is based on the default calculus, which makes apparent cases where a law – or the Catala implementation thereof – fails to account of a situation, either because a rule is lacking or because two rules of equal importance apply at the same time.

We present work in progress on CUTECat, a concolic testing engine for Catala programs. This engine aims to identify all of Catala domain-specific errors – and in particular errors related to the default calculus – while also yielding an output that can be easily understood, and therefore reviewed by lawyers.