diff --git a/src/MeshWeaver.Documentation/Data/WhatsNew/2026-08-26-openstreetmap-maps-actually-appear.md b/src/MeshWeaver.Documentation/Data/WhatsNew/2026-08-26-openstreetmap-maps-actually-appear.md
new file mode 100644
index 0000000000..a5783ac4a2
--- /dev/null
+++ b/src/MeshWeaver.Documentation/Data/WhatsNew/2026-08-26-openstreetmap-maps-actually-appear.md
@@ -0,0 +1,27 @@
+---
+Name: OpenStreetMap maps actually appear
+Category: Fix
+Description: OpenStreetMap maps are drawn at a usable size instead of collapsing to nothing, a map that cannot load explains itself on the page, and leaving a page mid-draw is no longer recorded as a failure.
+Icon: Globe
+Order: -20260826
+---
+
+# OpenStreetMap maps actually appear
+
+An OpenStreetMap map used to leave an empty rectangle on the page, and nothing said why. Three
+separate reasons could produce that same blank box, and none of them told the viewer anything.
+
+**The map now has a size.** A map is drawn into whatever space its container has, and that container
+had no height at all unless the page author happened to give it one — so a map that had loaded
+perfectly and was working correctly was still nothing to look at. It now comes with a sensible
+default height, and a page that wants a different one still just says so.
+
+**A map that cannot be drawn explains itself.** Where the map's own drawing code cannot be fetched,
+the page now shows the same short notice the portal already uses for any view that fails to render:
+what happened, a suggestion to reload, and the technical reason — including the exact address that
+could not be fetched, so the problem can be reported and fixed instead of guessed at.
+
+**Leaving the page is no longer a failure.** Navigating away while maps are still being drawn
+cancels that work, which is entirely routine — but it was being recorded with the same severity as
+a map that genuinely failed. Ordinary navigation is now recorded as what it is, which means what
+remains under that heading is worth looking at.