Record prune verification and State Hub sync latency evidence

Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a06ecb-456a-71c2-b41e-0755d336e883

docs/evidence/repo-review-2026-09-05.md

@@ -42,3 +42,28 @@ returned. No registrar skip was reported, so registrar authority was not
 assumed. Re-run `statehub fix-consistency` when these reads complete reliably
 to index the new workplans and refresh the generated records/brief. Progress
 was logged successfully as `7e317a96-eb7e-4b36-9d81-4efe49efa22c`.
+
+
+## Follow-up: prune guard and consistency latency
+
+The CLI now refuses apply before authentication if any explicitly requested
+image export is missing, unreadable, empty, or comment-only. Focused tests:
+14 passed. Full suite: 166 passed. No production prune was run.
+Implementation commit: `f637989`.
+
+Read-only HTTP timing reproduced the consistency problem:
+
+- `/repos/railiance-platform`: HTTP 200 in 13.58 seconds.
+- `/repos/railiance-platform/work-record-projection/snapshot`: HTTP 200 in
+  16.02 seconds; schema and repository ID match the checker contract.
+- The installed consistency checker defaults to a 10-second HTTP timeout;
+  snapshot failure causes fallback to individual workplan/task reads.
+
+A further ordinary sync was interrupted during slow reads. The installed
+checker was then invoked in-process with only its HTTP client timeout increased
+to 60 seconds (same `--here . --fix` arguments and repair/identity rules).
+It also failed to return a repair report within several minutes and was stopped;
+no generated repository changes were observed. This confirms a latency issue,
+but does not establish that timeout tuning alone is sufficient. State Hub sync
+remains pending. A focused State Hub endpoint/query latency investigation is
+needed before another full consistency run. No registrar authority was changed.