diff --git a/docs/LOGBOOK.md b/docs/LOGBOOK.md
index 871c96c3..0de584ba 100644
--- a/docs/LOGBOOK.md
+++ b/docs/LOGBOOK.md
@@ -1,3 +1,102 @@
+## 2026-05-19｜T85 Alert Chain durable metric evidence
+
+**背景**：
+
+- T84 後留下的技術債：`awoooi_alert_chain_last_success_timestamp` 原本只靠 API process-local Prometheus gauge。
+- 當 API pod 重啟或部署切換時，Prometheus 可能短暫看不到「最後告警鏈成功時間」，post-deploy smoke 也可能在新 webhook 進來前誤判缺資料。
+- 這會讓 Operator Console / Telegram 對「告警鏈到底有沒有正常跑完」產生不必要的疑慮。
+
+**完成變更**：
+
+- 新增 `apps/api/src/services/alert_chain_metrics_service.py`。
+- API `/metrics` 在輸出前會從 durable DB evidence 補回 `awoooi_alert_chain_last_success_timestamp`：
+  - `awooop_conversation_event`：`alertmanager / sentry / signoz` 的 internal timeline evidence。
+  - `alert_operation_log`：legacy Alertmanager receive evidence fallback。
+- 只補 `last_success_timestamp`，不補 `awoooi_alert_chain_healthy`，避免舊成功證據蓋掉新的 runtime failure。
+- 15 秒 in-process cache，避免 `/metrics` scrape 每次都打 DB。
+
+**驗證**：
+
+```text
+python3 -m py_compile apps/api/src/services/alert_chain_metrics_service.py apps/api/src/main.py apps/api/tests/test_alert_chain_metrics_service.py
+  -> OK
+
+PYTHONPATH=apps/api DATABASE_URL='postgresql+asyncpg://test:test@localhost/test' apps/api/.venv/bin/python -m pytest apps/api/tests/test_alert_chain_metrics_service.py -q
+  -> 2 passed
+
+PYTHONPATH=apps/api DATABASE_URL='postgresql+asyncpg://test:test@localhost/test' apps/api/.venv/bin/python -m pytest apps/api/tests/test_alert_chain_metrics_service.py apps/api/tests/test_adr100_slo_metrics_service.py -q
+  -> 5 passed
+
+apps/api/.venv/bin/python -m ruff check --select F,E9 apps/api/src/services/alert_chain_metrics_service.py apps/api/src/main.py apps/api/tests/test_alert_chain_metrics_service.py
+  -> All checks passed
+
+git diff --check
+  -> OK
+
+Pre-deploy production DB evidence:
+  -> alertmanager from awooop_conversation_event
+  -> alertmanager from alert_operation_log
+  -> sentry from awooop_conversation_event
+  -> signoz from awooop_conversation_event
+
+Commit:
+c516f9fc fix(metrics): refresh alert chain timestamp from durable evidence
+
+Gitea Actions:
+2472 Code Review for c516f9fc -> success
+2471 CD for c516f9fc -> success
+  tests -> success
+  build-and-deploy -> success
+  post-deploy-checks -> success
+
+Deploy marker:
+6f6cf90 chore(cd): deploy c516f9f [skip ci]
+
+Post-deploy evidence:
+  -> Alert Chain Metric 最後告警成功: 1 分鐘前
+  -> Alert Chain Smoke 8/8 checks passed
+  -> OTEL Collector 2 Pod(s) Running
+  -> Event Exporter 1 Pod(s) Running
+  -> Playwright smoke 5 passed
+  -> CI/CD success notification mirrored through AWOOI API
+```
+
+**Production smoke**：
+
+```text
+K8s image:
+awoooi-api    192.168.0.110:5000/awoooi/api:c516f9fc71f358de46d566625ef9c1eb164c102d
+awoooi-worker 192.168.0.110:5000/awoooi/api:c516f9fc71f358de46d566625ef9c1eb164c102d
+awoooi-web    192.168.0.110:5000/awoooi/web:c516f9fc71f358de46d566625ef9c1eb164c102d
+
+GET https://awoooi.wooo.work/api/v1/health
+  -> healthy, prod, mock_mode=false
+
+API pod internal GET /metrics:
+  -> awoooi_alert_chain_last_success_timestamp{source="alertmanager"} 1.779185257861278e+09
+  -> awoooi_alert_chain_last_success_timestamp{source="sentry"} 1.778679960070588e+09
+  -> awoooi_alert_chain_last_success_timestamp{source="signoz"} 1.778679960111112e+09
+```
+
+**注意事項 / 技術債**：
+
+- 公網 `https://awoooi.wooo.work/metrics` 目前會被 Web locale middleware 轉到 `/zh-TW/metrics`；真正 API metrics 應以 pod/service internal 或 Prometheus scrape 為準。
+- post-deploy log 仍會印出既有 `errSymlink cleanup noise` 註解，但 job 已 success，且沒有 T82 前的 runner panic。
+- 下一段可把 Alert Chain durable evidence 也映射到前端 AwoooP Runs / Monitoring 的「證據來源」欄位，讓 operator 看得到此指標是來自 DB evidence，不只是 Prometheus scrape。
+
+**目前整體進度**：
+
+- AwoooP 告警可觀測鏈：約 97.8%。
+- 低風險自動修復閉環：約 95%。
+- 前端 AI 自動化管理介面同步：約 92.5%。
+- CI/CD notification AwoooP 主路徑：約 99%。
+- CI/CD runner hygiene：約 96%。
+- 治理告警可讀性 / 可處置性：約 90%。
+- Alert Chain durable metric evidence：約 92%。
+- 完整 AI 自動化管理產品化：約 90.0%。
+
+---
+
 ## 2026-05-19｜T82 Gitea CD E2E smoke symlink cleanup hygiene
 
 **背景**：