diff --git a/.ci/linters/md/news_linter.R b/.ci/linters/md/news_linter.R
new file mode 100644
index 000000000..5eeb302a3
--- /dev/null
+++ b/.ci/linters/md/news_linter.R
@@ -0,0 +1,61 @@
+any_mismatch = FALSE
+
+# ensure that numbered list in each section is in sequence
+check_section_numbering = function(news) {
+  # plain '#' catches some examples
+  sections = grep("^#+ [A-Z]", news)
+  entries = grep("^[0-9]+[.]", news)
+  entry_value = as.integer(gsub("^([0-9]+)[.].*", "\\1", news[entries]))
+  section_id = findInterval(entries, sections)
+  
+  any_mismatch = FALSE
+  for (id in unique(section_id)) {
+    section_entries = entry_value[section_id == id]
+    intended_value = seq_along(section_entries)
+    matched = section_entries == intended_value
+    if (all(matched)) next
+    any_mismatch = TRUE
+    section_header = news[sections[id]]
+    cat(sprintf(
+      "In section '%s' (line %d), bad numbering:\n%s\n",
+      section_header, sections[id],
+      paste0("  [", section_entries[!matched], " --> ", intended_value[!matched], "]", collapse="\n")
+    ))
+  }
+  return(any_mismatch)
+}
+
+# ensure that GitHub link text & URL actually agree
+check_gh_links = function(news) {
+  gh_links_info = gregexpr(
+    "\\[#(?<md_number>[0-9]+)\\]\\(https://github.com/Rdatatable/data.table/(?<link_type>[^/]+)/(?<link_number>[0-9]+)\\)",
+    news,
+    perl=TRUE # required for within-group indices
+  )
+  gh_link_metadata = do.call(rbind, lapply(seq_along(gh_links_info), function(idx) {
+    x = gh_links_info[[idx]]
+    if (x[1L] <= 0L) return(NULL)
+    match_mat = attr(x, "capture.start") # matrix seeded with the correct dimensions
+    match_mat[] = substring(news[idx], match_mat, match_mat + attr(x, "capture.length") - 1L)
+    match_df = data.frame(match_mat)
+    match_df$line_number = idx
+    match_df
+  }))
+  matched = gh_link_metadata$md_number == gh_link_metadata$link_number
+  if (all(matched)) return(FALSE)
+
+  cat(sep = "", with(gh_link_metadata[!matched, ], sprintf(
+    "In line %d, link pointing to %s %s is written #%s\n",
+    line_number, link_type, link_number, md_number
+  )))
+  return(TRUE)
+}
+
+any_error = FALSE
+for (news in list.files(pattern = "NEWS")) {
+  cat(sprintf("Checking NEWS file %s...\n", news))
+  news_lines = readLines("NEWS.md")
+  any_error = any_error || check_section_numbering(news_lines)
+  any_error = any_error || check_gh_links(news_lines)
+}
+if (any_error) stop("Please fix the NEWS issues above.")
diff --git a/.github/workflows/code-quality.yaml b/.github/workflows/code-quality.yaml
index b38a7929c..21b416dc7 100644
--- a/.github/workflows/code-quality.yaml
+++ b/.github/workflows/code-quality.yaml
@@ -70,3 +70,8 @@ jobs:
           }
           cat("All translation quality checks completed successfully!\n")
         shell: Rscript {0}
+
+  lint-md:
+      - name: Lint
+        run: |
+          for (f in list.files('.ci/linters/md', full.names=TRUE)) source(f)
diff --git a/NEWS.md b/NEWS.md
index 29047f1a1..c7aec62b1 100644
--- a/NEWS.md
+++ b/NEWS.md
@@ -351,7 +351,7 @@
 
 23. `DT[, head(.SD,n), by=grp]` and `tail` are now optimized when `n>1`, [#5060](https://github.com/Rdatatable/data.table/issues/5060) [#523](https://github.com/Rdatatable/data.table/issues/523#issuecomment-162934391). `n==1` was already optimized. Thanks to Jan Gorecki and Michael Young for requesting, and Benjamin Schwendinger for the PR.
 
-24. `setcolorder()` gains `before=` and `after=`, [#4385](https://github.com/Rdatatable/data.table/issues/4358). Thanks to Matthias Gomolka for the request, and both Benjamin Schwendinger and Xianghui Dong for implementing. Also thanks to Manuel López-Ibáñez for testing dev and mentioning needed documentation before release.
+24. `setcolorder()` gains `before=` and `after=`, [#4358](https://github.com/Rdatatable/data.table/issues/4358). Thanks to Matthias Gomolka for the request, and both Benjamin Schwendinger and Xianghui Dong for implementing. Also thanks to Manuel López-Ibáñez for testing dev and mentioning needed documentation before release.
 
 25. `base::droplevels()` gains a fast method for `data.table`, [#647](https://github.com/Rdatatable/data.table/issues/647). Thanks to Steve Lianoglou for requesting, Boniface Kamgang and Martin Binder for testing, and Jan Gorecki and Benjamin Schwendinger for the PR. `fdroplevels()` for use on vectors has also been added.