I agree. I think it may be a little bit less straightfoward than the other 2 (use_r() and use_test())

Should there be an option for a generic test helper "tests/testthat/helper.R", vs file specific test helper?

Also, important to note that R/helper.R and tests/testthat/helper.R may not be related.

I made a little sketch. I don't really like that the default isn't consistent with the default of use_test(), but also I think that having a helper file for each test file by default seems a bit overkill.

#' @rdname use_r
#' @export
#' @param scope if global, creates "tests/testthat/helper.R", otherwise creates
#'   "tests/testthat/helper-{name}.R
use_test_helper <- function(name = NULL, scope = c("global", "file"), open = rlang::is_interactive()) {
  if (!uses_testthat()) {
    use_testthat_impl()
  }
  scope <- arg_match(scope)

  if (scope == "global") {
    path <- path("tests", "testthat", "helper", ext = "R")
  } else {
    path <- path("tests", "testthat", paste0("helper-", compute_name(name)))
  }

  edit_file(proj_path(path), open = open)
  ui_bullets(c(
    "_" = "Run {.run devtools::load_all()} to load objects from helper files in
         your environment."
  )
  )
  invisible(TRUE)
}

I'm generally on board, so am happy to see this in PR form. I'm not sure the name and scope arguments are quite right yet. I don't have great vocabulary for stuff yet, but clearly we want to make it easy to ask for helper.R versus helper-stuff.R. I think 1 argument is enough. "Suffix" almost captures it, but it's not perfect 🤔