refactor: retain decision state instead of full logs

docs/failure-analysis-2026-09-04.md

@@ -68,16 +68,20 @@ The most useful structured codes were `MODEL_NOT_SUPPORTED` (1,700),
 
 ## Durable-state corrections
 
-- Hot intent history retains unresolved intents plus 200 recent terminal
-  intents; older terminal attempts are gzip archived by month.
+- Hot intent history retains unresolved intents plus 300 compact recent
+  terminal intents; older terminal attempts are discarded after extraction.
 - Community raw samples are capped locally at 200 and excluded from the hot Git
   snapshot. The aggregated GPU/framework statistics remain durable.
-- Model/GPU official-capability cache is bounded to the 1,500 newest entries.
-- Outcome compaction starts at 1,000 rows instead of 2,000.
+- Model/GPU official-capability cache is bounded to the 750 newest entries.
+- Outcome compaction starts at 500 rows and retains 300 compact recent samples.
 - An unchanged snapshot produces no Git commit. A failed push retries the exact
   same commit and generation instead of creating a new generation every minute.
 - Submission intent batches default to 100, reducing Git transactions while
   preserving write-ahead recovery.
 
-These changes keep full forensic evidence in cold archive branches while making
-the hot branch small enough for quick restart and reliable server-side unpack.
+After this audit was completed, version `2026.09.04.2` changed the ongoing
+retention model to `decision_state_only`. The extracted aggregate statistics,
+compatibility rules, memory observations, active recovery state, and compact
+recent samples remain durable, but new full outcome/intent archives and the
+duplicated event stream are no longer produced. Previously created archive
+branches remain untouched and are not needed during startup.