Jev in Practice: A Composable Python Toolkit for TypeSafe's System One Decision Model

daf-jev 0.7.1 provides typed decision clients, provider-neutral backends, pure composition, exact discrete graphical inference, frozen benchmark protocols, and reproducible manuscript evidence. This software release adds an explicit shared USD 25 hosted allocation, decimal liability reservations, immutable imports of historical attempts, and fail-closed accounting. The retained hosted pilot has one failed capability attempt with unresolved billing and no hosted quality comparison; the full comparative study remains unfinished. Published evidence distinguishes CPU comparators, partial historical native-model results, and current source-level verification. Source, wheel, sdist, 28-page manuscript, selected evidence archive, release manifest, and checksums are the exact assets of the independently verified GitHub v0.7.1 release. No credentials, private journals, model weights, or raw corpora are included.