AutoCause: Python Framework Automates Environmental Causal Discovery.

Environmental time-series causal discovery often relies on inconsistent expert decisions regarding method selection, statistical tests, and interpretation, leading to non-comparable or irreproducible results. A new open-source Python framework, AutoCause, addresses this by automating these expert choices. It records each decision, provides default settings derived from an extended causal-audit module, and allows for domain-informed overrides. AutoCause integrates four established causal discovery methods from different families, includes non-causal reference models, and grades causal links based on method support. Benchmarking on 145 datasets showed that the methods recover complementary parts of reference graphs, with majority-supported links demonstrating higher precision on synthetic data. This framework transforms inconsistent expert practices into an auditable and repeatable analysis, though causal interpretation still rests with the analyst.