diff --git a/.gitignore b/.gitignore
index ad0f893..3f2fb09 100644
--- a/.gitignore
+++ b/.gitignore
@@ -126,3 +126,9 @@ public/
 # but this file changes on every cycle, and committing it would mean a commit
 # per run saying nothing changed.
 snapshots/*.state.json
+
+# Half-written snapshots. Every snapshot is written to a sibling temp file and
+# renamed into place, so a run killed mid-write leaves one of these behind — and
+# the refresh workflow commits the whole snapshots/ directory, which would
+# otherwise pin a truncated page in git forever.
+snapshots/*.tmp-*
diff --git a/snapshots/README.md b/snapshots/README.md
index f0dbea4..adc9ccc 100644
--- a/snapshots/README.md
+++ b/snapshots/README.md
@@ -17,7 +17,8 @@ localStorage keys).
 
 Every file is written to a sibling `.tmp-*` and renamed into place, body before metadata, so an
 interrupted run leaves a stray temp file rather than a truncated snapshot or metadata describing
-bytes that were never stored.
+bytes that were never stored. Those temp files are gitignored: `refresh.yml` commits the whole
+directory, so otherwise a run killed mid-write would pin a half-page in git forever.
 
 Three reasons this is committed rather than cached: