diff --git a/.github/workflows/code-test.yaml b/.github/workflows/code-test.yaml
index a6fea538d..bbfc0e2f8 100644
--- a/.github/workflows/code-test.yaml
+++ b/.github/workflows/code-test.yaml
@@ -93,12 +93,27 @@ jobs:
     needs: [uv]
     runs-on: ubuntu-latest
     steps:
+      - uses: actions/checkout@d23441a48e516b6c34aea4fa41551a30e30af803 # v6
+      # No merge-multiple: each project's coverage.xml has the same filename, so
+      # flattening would overwrite them. Download each artifact into its own
+      # coverage-reports/<artifact-name>/ subdirectory instead.
       - uses: actions/download-artifact@9000827ccba6bdab643e8b6fd33ac0654aef8333 # v8
         with:
           pattern: coverage-*
-          merge-multiple: true
+          path: coverage-reports
+      # Merge every project's coverage.xml into one report. This used to call
+      # the `cobertura-merge-globby` npm package, which uses `util.isString()`
+      # (removed in Node 24) and now crashes on the runner. A stdlib Python
+      # script does the same job with no npm/Node dependency.
       - run: |
-          npx cobertura-merge-globby -o output.xml --files=python/**/coverage.xml
+          shopt -s globstar nullglob
+          files=(coverage-reports/**/coverage.xml)
+          if [ ${#files[@]} -eq 0 ]; then
+            echo "No coverage.xml artifacts found to merge" >&2
+            exit 1
+          fi
+          python3 scripts/merge-cobertura.py -o coverage.xml "${files[@]}"
+        shell: bash
       - uses: actions/upload-artifact@cf430e030ddbb5b0abf93d22962f4752f3646cd9 # v7
         with:
           name: coverage.xml
diff --git a/scripts/merge-cobertura.py b/scripts/merge-cobertura.py
new file mode 100755
index 000000000..f08156543
--- /dev/null
+++ b/scripts/merge-cobertura.py
@@ -0,0 +1,123 @@
+#!/usr/bin/env python3
+"""Merge multiple Cobertura coverage XML files into a single report.
+
+Replaces the unmaintained `cobertura-merge-globby` npm package, which calls
+`util.isString()` -- a Node API removed in Node 24 -- and crashes on the
+current GitHub Actions runner image.
+
+Each input is a Cobertura `<coverage>` document whose real content lives under
+`<packages>`. Merging concatenates every input's `<package>` elements into one
+output document and recomputes the top-level line-rate / branch-rate from the
+combined hit counts, which is what downstream consumers (e.g. orgoro/coverage)
+read. Inputs that are missing or unparsable are skipped with a warning rather
+than aborting the merge -- some projects (stestr-based) never emit a
+coverage.xml, so the set of inputs is expected to vary.
+"""
+
+from __future__ import annotations
+
+import argparse
+import sys
+import xml.etree.ElementTree as ET
+from pathlib import Path
+
+
+def _rate(covered: int, total: int) -> float:
+    return covered / total if total else 0.0
+
+
+def merge(paths: list[Path], output: Path) -> int:
+    merged_packages: list[ET.Element] = []
+    sources: list[str] = []
+
+    lines_covered = lines_valid = 0
+    branches_covered = branches_valid = 0
+    parsed = 0
+
+    for path in paths:
+        if not path.is_file():
+            print(f"warning: skipping missing coverage file: {path}", file=sys.stderr)
+            continue
+        try:
+            root = ET.parse(path).getroot()
+        except ET.ParseError as exc:
+            print(f"warning: skipping unparsable {path}: {exc}", file=sys.stderr)
+            continue
+
+        parsed += 1
+
+        # Cobertura stores per-line/branch totals as attributes on <coverage>.
+        # Accumulate them so the merged top-level rates stay accurate.
+        lines_covered += int(root.get("lines-covered", 0))
+        lines_valid += int(root.get("lines-valid", 0))
+        branches_covered += int(root.get("branches-covered", 0))
+        branches_valid += int(root.get("branches-valid", 0))
+
+        for sources_el in root.findall("sources"):
+            for source in sources_el.findall("source"):
+                if source.text and source.text not in sources:
+                    sources.append(source.text)
+
+        for packages_el in root.findall("packages"):
+            merged_packages.extend(packages_el.findall("package"))
+
+    if parsed == 0:
+        print(
+            "error: no coverage files could be parsed; nothing to merge",
+            file=sys.stderr,
+        )
+        return 1
+
+    coverage = ET.Element("coverage")
+    coverage.set("line-rate", f"{_rate(lines_covered, lines_valid):.4f}")
+    coverage.set("branch-rate", f"{_rate(branches_covered, branches_valid):.4f}")
+    coverage.set("lines-covered", str(lines_covered))
+    coverage.set("lines-valid", str(lines_valid))
+    coverage.set("branches-covered", str(branches_covered))
+    coverage.set("branches-valid", str(branches_valid))
+    coverage.set("version", "1.0")
+    coverage.set("timestamp", "0")
+
+    if sources:
+        sources_el = ET.SubElement(coverage, "sources")
+        for src in sources:
+            ET.SubElement(sources_el, "source").text = src
+
+    packages_out = ET.SubElement(coverage, "packages")
+    for package in merged_packages:
+        packages_out.append(package)
+
+    tree = ET.ElementTree(coverage)
+    ET.indent(tree, space="  ")
+    output.write_bytes(
+        b'<?xml version="1.0" ?>\n' + ET.tostring(coverage, encoding="unicode").encode()
+    )
+    print(
+        f"merged {parsed} coverage file(s), "
+        f"{len(merged_packages)} package(s) -> {output}",
+        file=sys.stderr,
+    )
+    return 0
+
+
+def main(argv: list[str] | None = None) -> int:
+    parser = argparse.ArgumentParser(description=__doc__)
+    parser.add_argument(
+        "-o",
+        "--output",
+        type=Path,
+        required=True,
+        help="path to write the merged coverage XML",
+    )
+    parser.add_argument(
+        "inputs",
+        nargs="+",
+        type=Path,
+        help="Cobertura coverage XML files to merge",
+    )
+    args = parser.parse_args(argv)
+    return merge(args.inputs, args.output)
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())