The documentation for that package is absolutely terrible (please feel encouraged to contact the author and complain...when a paper over an R/other software package is in review, we really need to include the documentation/tutorial in the review process), so I'm surprised you're stumped. It turns out that you have to look through the source code to see what's actually used from genepos and what things need to be named as.

Assuming you have a reasonably well formatted GTF for your organism (i.e., not from UCSC), then something like the following R code should work to create genepos.

library(rtracklayer)
gtf <- import.gff("Mus_musculus.GRCm38.71.gtf", format="gtf")

stupidNASTIseqFunction <- function(gtf) {
    #compact each gene to a single entry
    grl <- split(gtf, gtf$gene_id)
    #This won't work on the weird GTF files from UCSC
    res <- lapply(grl, function(x) {
        start(x)[1] = min(start(x))
        end(x)[1] = max(end(x))
        x[1,]
    })
    #format the output data frame
    df <- data.frame(
        seqname = factor(sapply(res, function(x) as.character(seqnames(x)))),
        start = sapply(res, start),
        end = sapply(res, end),
        strand = factor(sapply(res, function(x) as.character(strand(x)))),
        attributes=sapply(res, function(x) x$gene_id))
    #reorder things, since grl is ordered be gene_id
    o <- match(unique(gtf$gene_id), df$attribute)
    df[o,]
}

genepos <- stupidNASTIseqFunction(gtf)

You'll need to change the import.gff(...) part to match your own GTF file. For what it's worth, this method is rather slow with large GTF files, but you'll only need to do it once.