Ecosystem update — September 2026

Related modelling addition elsewhere in the suite

The related gpbiometricspy package now includes a fully exact-main-certified crossed participant–item Gaussian hierarchical location–scale model with one location random slope for each crossed factor.

This is an ecosystem note only. It does not change GazeAudit’s audit logic, release record, validation evidence or scientific claims. The gpbiometricspy addition is a separate modelling path for continuous repeated outcomes with crossed participant/item heterogeneity.

Certification record

Certified gpbiometricspy PR #129 is pinned to merge SHA d078e0366ace49c3ebeb2f6800bad6394d70631e:

• 14/14 exact-main push workflow families green;
• 12/12 Ubuntu/macOS/Windows × Python 3.11–3.14 test lanes green;
• 782/782 tests;
• 14,015/14,015 statements;
• 6,757/6,776 raw branches = 99.7196%;
• 19 unchanged audited structural arcs with 0 unexpected, 0 stale and 0 unaudited debt;
• frozen gpbiometrics 2.0.0 parity unchanged at 406/406.

Read the crossed participant–item random-slope guide

Open gpbiometricspy PR #129

GazeAudit currently has repository documentation but no root MkDocs site configuration; this page updates the GitHub documentation source without inventing a new Pages/deployment pipeline.