diff --git a/CHANGELOG.txt b/CHANGELOG.txt
index cb16688b..45bf1c13 100644
--- a/CHANGELOG.txt
+++ b/CHANGELOG.txt
@@ -1,4 +1,26 @@
 2026-03-05
+perf: add organization_id to alerts for fast single-table queries
+  - Denormalize organization_id from device to alerts table for query performance
+  - Add composite indexes on (organization_id, alert_type, resolved_at)
+  - Optimize list_organization_active_alerts to eliminate expensive 3-table joins
+  - Auto-populate organization_id when creating alerts
+Files: lib/towerops/alerts/alert.ex, alerts.ex
+      priv/repo/migrations/20260305152930_add_organization_id_to_alerts.exs
+
+fix: change agent token cascade deletes to RESTRICT for data safety
+  - Changed agent_tokens.organization_id from CASCADE to RESTRICT on delete
+  - Changed agent_assignments.agent_token_id from CASCADE to RESTRICT on delete
+  - Prevents silent data loss when deleting organizations or agent tokens
+Files: priv/repo/migrations/20260305152524_fix_agent_cascade_deletes.exs
+
+fix: add validation for state_descr in sensor readings
+  - Added min/max length validation (1-255 chars) for state_descr field
+  - Prevents empty or overly long state descriptions from being stored
+Files: lib/towerops/snmp/sensor_reading.ex
+
+ci: upgrade to Elixir 1.19.5 and OTP 28.3 to match .tool-versions
+Files: .forgejo/workflows/build.yaml
+
 fix: comprehensive reliability audit bug fixes (16 bugs fixed)
   - Critical (5): Task.yield_many race condition, Enum.zip data corruption, missing Alert schema fields, LiveView timer/PubSub leaks
   - High (3): clock skew handling, nil crash in interface status, migration index naming
diff --git a/priv/static/changelog.txt b/priv/static/changelog.txt
index fd3aa548..930474f5 100644
--- a/priv/static/changelog.txt
+++ b/priv/static/changelog.txt
@@ -1,4 +1,7 @@
 2026-03-05 — Reliability & Performance Improvements
+* Optimized alert queries for faster dashboard loading on large organizations
+* Improved data integrity with stricter database constraints
+* Enhanced validation for sensor readings to prevent invalid data
 * Fixed memory leaks in real-time dashboard components
 * Improved error handling for device polling and monitoring
 * Better handling of clock synchronization issues between servers