causal-learn
causal-learn is an open-source Python causal-discovery toolkit that helps researchers and data scientists learn candidate causal structures from data and run related analyses.
Tool overview
Adoption judgment: causal-learn is worth considering for research or prototyping when you know Python, statistics, and the assumptions behind causal discovery and need to run discovery algorithms on data. It should not be adopted on the assumption that it is a turnkey system that validates causal claims or produces business-ready conclusions without expert review.
More precisely, it is a Python algorithm toolkit for exploring candidate causal structures, not a general machine-learning platform, a standalone visualization product, or a causal-effect and decision-making service. The available description indicates support for multiple causal-discovery algorithms and related analysis workflows, with a structure or graph-like result as the main output. Such results still depend on data quality, variable selection, model assumptions, and follow-up validation; a discovered structure is not automatically proof of a causal relationship established by an experiment.
The cost evidence is limited. A community article demonstrates installation with pip, but that is a community usage example rather than official pricing or a service-level promise.