There were 0 errors, 0 warnings, and 0 notes.

• macOS Tahoe 26.5.1 (aarch64), R 4.6.1 (2026-06-24), checked locally
• win-builder (R-devel) — see pre-submission checklist below
• R-hub (Ubuntu, R-release) — see pre-submission checklist below

This is the first CRAN submission of bsync. There are no reverse dependencies.

• The package uses compiled C++ code (Rcpp/RcppArmadillo) for the windowed estimator cores. The compiled code is serial (no OpenMP), reproducible, and has been tested with valgrind on the local check.

• Vignette build times: the most expensive vignette (choosing-parameters) runs a surrogate analysis with n_surrogates = 100 on sim_dyad (a bundled synthetic dataset of 2,400 rows). All six vignettes build in under 2 minutes total on the test machine. Tests marked skip_on_cran() are the heavy surrogate runs (1,000+ permutations).

• The sim_dyad dataset is a small synthetic dataset (~250 KB) generated from a known ground-truth model for use in examples and vignettes.

• Deploy pkgdown site: the site is deployed; the articles/bsync.html URL referenced in README is live (HTTP 200) and urlchecker::url_check() reports all URLs correct.

• urlchecker clean: urlchecker::url_check() — all URLs correct.

• spelling clean: spelling::spell_check_package() — no errors.

• pkgdown clean: pkgdown::check_pkgdown() — no problems found.

• win-builder check: submit to devtools::check_win_devel() and devtools::check_win_release(); confirm 0/0/0.

• R-hub check: run rhub::rhub_check() on at least Ubuntu + Windows builders; confirm 0/0/0.

• Final local check: devtools::check(args = '--as-cran') = 0/0/0.

• Submit via devtools::submit_cran().