I enabled all kinds of analyzes in the config file that I found, and it still works (on my machine 😅) => ready for review and merge

jfi: I also played with codex to automatically configure the workflow with prompts like i only need PCA, can you disable the rest? only edit the config file and it worked pretty well 🔥

merged assuming complete testing and only linting related changes

I enabled all kinds of analyzes in the config file that I found, and it still works (on my machine 😅) => ready for review and merge

This contradicts your statement about the ClusterCrit package not being available for macOS. So either it is available and you tested its functionality, or it is not available and you couldn't have tested it. Please comment.

Also, I quickly checked, and at least on conda-forge, there is a macOS version (actually all three platforms are supported). Please comment as well on that.
https://anaconda.org/channels/conda-forge/packages/r-clustercrit/overview

1)

Codex used a dirty workaround making it working on my machine, it added an inline library fetch in validation_internal.R:

ensure_clusterCrit <- function() {
  if (requireNamespace("clusterCrit", quietly = TRUE)) {
    return(invisible(TRUE))
  }

  primary_lib <- .libPaths()[1]
  lock_dir <- file.path(primary_lib, "00LOCK-clusterCrit-codex")

  while (!dir.create(lock_dir, showWarnings = FALSE)) {
    if (requireNamespace("clusterCrit", quietly = TRUE)) {
      return(invisible(TRUE))
    }
    Sys.sleep(2)
  }

  on.exit(unlink(lock_dir, recursive = TRUE, force = TRUE), add = TRUE)

  install.packages(
    "clusterCrit",
    repos = "https://cloud.r-project.org",
    lib = primary_lib,
    quiet = TRUE
  )

  if (!requireNamespace("clusterCrit", quietly = TRUE)) {
    stop("Failed to install the CRAN package 'clusterCrit'.")
  }
}

ensure_clusterCrit()

But via conda envs it wasn't/isn't possible.

2)

Yeah, it is available on macOS (as well as Linux and Windows), but only for x86-64 CPUs. All new MacBooks use the ARM64 architecture, that’s the problem.

Okay, understood. Thanks.