diff --git a/docs/LOGBOOK.md b/docs/LOGBOOK.md
index 5e1246b2..60aa16ac 100644
--- a/docs/LOGBOOK.md
+++ b/docs/LOGBOOK.md
@@ -72,6 +72,24 @@
 - Monitoring / alert / PlayBook / Telegram event packet / live scrape：`100%`。
 - Runtime remediation / Telegram 實發 / Bot API call / host write：仍 `0 / false`；本段未發 Telegram、未讀 secret、未 kill process、未重啟服務、未改 firewall/K8s。
 
+### 2026-06-18 14:51 台北｜Host runaway event packet 正式部署 / production readback
+
+**部署錨點**：`f358a0f6 fix(api): route runaway host alerts to ai event packets` 已由 Gitea CD `#3150` 部署，deploy marker 為 `2d278568 chore(cd): deploy f358a0f [skip ci]`。CD 證據顯示 tests job `3221 passed, 23 skipped`、B5 integration `5 passed`、API / Web image 皆使用 `f358a0f6c3e614e407dedb6eee89bf10b2bc8173`，post-deploy alert-chain smoke `9/9` 通過，monitoring coverage `14/14` jobs up。
+
+**Runtime readback**：
+- ArgoCD `awoooi-prod`：`Synced / Healthy`，revision `2d278568cb60e1bd79af4275249df374f220039f`。
+- Kubernetes deployments：`awoooi-api 2/2`、`awoooi-web 2/2`、`awoooi-worker 1/1`，image revision 均為 `f358a0f6c3e614e407dedb6eee89bf10b2bc8173`。
+- Production health：`https://awoooi.wooo.work/api/v1/health` 回 `healthy / prod / mock_mode=false`。
+- Production routes：`/zh-TW/iwooos`、`/zh-TW/governance?tab=automation-inventory`、`/zh-TW/awooop/tenants` 皆 `200`；抽樣掃描 `codex_delegation`、`source_thread_id`、`批准！`、`192.168.0.110`、`192.168.0.120` 命中 `0`。
+- Prometheus：`awoooi_host_runaway_process_monitor_up{host="110"}=1`，兩條 orphan browser group count 均為 `0`，`awoooi_host_gitea_actions_active_container_count{host="110"}=2`，`awoooi_host_runaway_process_remediation_authorized{host="110"}=0`；`HostRunawayProcessMonitorMissing` 與 `HostOrphanBrowserSmokeHighCpu` 未 firing。
+
+**完成度同步**：
+- 110 CPU runaway observe -> classify -> alert -> AI event packet -> PlayBook / KM contract：`100%` 並已正式部署讀回。
+- Live observability：`100%`，110 textfile exporter 已安裝、Prometheus 已 scrape。
+- Runtime remediation：仍 `0% / false`，這是刻意保留的安全 gate；沒有 owner approval、maintenance window、evidence ref、dry-run 與 post-check 時，不允許自動 kill、Docker restart、systemd restart、Nginx reload、firewall change 或 K8s action。
+
+**下一步**：若未來 `HostOrphanBrowserSmokeHighCpu` 真正 firing，處理順序必須是 AI triage packet -> remediation dry-run evidence -> owner approval / maintenance window / evidence ref -> gated `SIGTERM` -> post-check -> KM / PlayBook / Verifier 回寫；合法 CI load 則走容量 / 排程 / runner queue 分流，不走 process remediation。
+
 ## 2026-06-18｜P2-406B Receipt Readback Owner Review 本地完成
 
 **背景**：P2-004 已把依賴 / 供應鏈漂移收斂成只讀監控讀回；統帥要求每次推進都不能忘記目標與方向，因此本段把日報 / 週報 / 月報、Telegram receipt owner review、P2-004 drift monitor 與 P2-403J 報表真相串成同一個 owner review surface，讓治理頁可以直接看到 AI Agent 分工、互審與仍被關閉的 runtime 邊界。
diff --git a/docs/runbooks/FULL-STACK-COLD-START-SOP.md b/docs/runbooks/FULL-STACK-COLD-START-SOP.md
index 5a17e692..47e24cf1 100644
--- a/docs/runbooks/FULL-STACK-COLD-START-SOP.md
+++ b/docs/runbooks/FULL-STACK-COLD-START-SOP.md
@@ -44,6 +44,18 @@ Allowed declaration: runaway Chrome/smoke recurrence guard is live and scraped.
 Forbidden declaration: AI runtime remediation is enabled. Remediation remains gated and must not execute without owner approval, maintenance window, evidence ref, dry-run, and post-check.
 ```
 
+2026-06-18 14:51 production event-packet readback:
+
+```text
+Host runaway alert-to-event packet is deployed in production.
+Deploy marker: 2d278568 chore(cd): deploy f358a0f [skip ci].
+Runtime image: awoooi-api / awoooi-web / awoooi-worker use f358a0f6c3e614e407dedb6eee89bf10b2bc8173.
+ArgoCD readback: awoooi-prod Synced / Healthy.
+Alert mapping: HostOrphanBrowserSmokeHighCpu -> orphan_browser_smoke_runaway_process; HostCiRunnerLoadSaturation -> ci_runner_load_saturation.
+Allowed declaration: monitoring, alert rules, live scrape, AI event packet routing, PlayBook / KM contract, and production deployment are complete for this evidence set.
+Forbidden declaration: Telegram send, Bot API call, Gateway queue write, process kill, Docker/systemd restart, Nginx reload, firewall/K8s action, or runtime remediation is authorized.
+```
+
 | 項目 | 2026-06-18 13:43 Asia/Taipei live result | 判定 |
 |------|-------------------------------------------|------|
 | Overall recovery readiness | `99%` | `SERVICE_GREEN_DR_ESCROW_BLOCKED` |
diff --git a/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md b/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
index 61d47eea..0f97285b 100644
--- a/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
+++ b/docs/superpowers/specs/2026-04-15-MASTER-ai-autonomous-flywheel-v2.md
@@ -5101,3 +5101,17 @@ Trigger commit `f5cd37b7` 與 deploy marker `0ba92357` 已把 governance UI 的
 - 精準測試 `apps/api/tests/test_telegram_message_templates.py` 已新增兩條 regression，`59 passed`；`telegram_gateway.py` py_compile 通過。
 
 **裁決：** 這是 alert -> AI event packet 的只讀與訊息模板閉環，不是 Telegram 實發、Bot API call、Gateway queue write、host write 或 process kill 授權。所有卡片仍固定 `runtime_write_gate=0`，真正修復仍必須走 dry-run、owner approval、maintenance window、evidence ref、post-check 與 KM / PlayBook 回寫。
+
+### 2026-06-18 14:51 (台北) — §8 / Host CPU AIOps — event packet 正式部署讀回
+
+**觸發**：`HostOrphanBrowserSmokeHighCpu` / `HostCiRunnerLoadSaturation` 專屬 lane 已在 source 與 regression test 完成，必須補 production 真相，避免把本地訊息模板完成誤判為正式站已可用。
+
+**已推進：**
+- `f358a0f6 fix(api): route runaway host alerts to ai event packets` 已由 Gitea CD `#3150` 正式部署，deploy marker `2d278568 chore(cd): deploy f358a0f [skip ci]`。
+- CD tests job 顯示 `3221 passed, 23 skipped`，B5 integration `5 passed`；post-deploy alert-chain smoke `9/9` 通過，monitoring coverage `14/14` jobs up。
+- ArgoCD `awoooi-prod` 讀回 `Synced / Healthy`，revision `2d278568cb60e1bd79af4275249df374f220039f`。
+- `awoooi-api`、`awoooi-web`、`awoooi-worker` 均已 rollout 到 image `f358a0f6c3e614e407dedb6eee89bf10b2bc8173`。
+- Production health 與 `/zh-TW/iwooos`、`/zh-TW/governance?tab=automation-inventory`、`/zh-TW/awooop/tenants` route smoke 通過；抽樣禁字命中 `0`。
+- Prometheus 仍讀到 110 runaway exporter：`monitor_up=1`、orphan browser group count `0`、active CI containers `2`、`remediation_authorized=0`，missing / orphan alerts 未 firing。
+
+**裁決：** 110 CPU runaway 的 observe -> classify -> alert -> AI event packet -> PlayBook / KM contract 已經正式站可用。這仍不是 runtime 修復授權：Telegram 實發、Bot API call、Gateway queue write、host write、process kill、Docker / systemd restart、Nginx reload、firewall / K8s action 仍維持 `0 / false`。未來若 alert firing，AI 只能先產生 triage packet、dry-run evidence、owner / maintenance / evidence gate 與 post-check 計畫，再由 gated remediation helper 執行最小 `SIGTERM`。
diff --git a/docs/workplans/2026-06-04-reboot-cold-start-backup-recovery-workplan.md b/docs/workplans/2026-06-04-reboot-cold-start-backup-recovery-workplan.md
index fa7b6e74..aec54f80 100644
--- a/docs/workplans/2026-06-04-reboot-cold-start-backup-recovery-workplan.md
+++ b/docs/workplans/2026-06-04-reboot-cold-start-backup-recovery-workplan.md
@@ -236,6 +236,15 @@ Evidence: `apps/api/src/services/telegram_gateway.py`、`apps/api/tests/test_tel
 Blocked: No for alert-to-event packet; yes for Telegram live send / runtime remediation by design.
 Next: 等 code-review / CD 後做 production readback；若未來 alert 實際 firing，確認 Telegram card 與 AwoooP Run truth-chain 都能呈現同一 lane。
 
+2026-06-18 14:51 Asia/Taipei
+Phase: P3 AI Ops alert-to-event packet production readback
+Before: `HostOrphanBrowserSmokeHighCpu` / `HostCiRunnerLoadSaturation` 已有 source + test，但尚未完成正式站部署與 runtime revision 讀回。
+After: `f358a0f6` 已由 Gitea CD `#3150` 部署，deploy marker `2d278568`。ArgoCD `awoooi-prod` 為 `Synced / Healthy`，API / Web / Worker image 均為 `f358a0f6c3e614e407dedb6eee89bf10b2bc8173`；production health 與 IwoooS / Governance / AwoooP tenants routes 皆 `200` 且敏感字串抽樣命中 `0`。
+Evidence: Gitea CD `#3150` tests `3221 passed, 23 skipped`、B5 integration `5 passed`、post-deploy alert-chain smoke `9/9`、monitoring coverage `14/14` jobs up；Prometheus 仍讀到 110 `monitor_up=1`、orphan browser group count `0`、CI active containers `2`、`remediation_authorized=0`，missing / orphan alerts 未 firing。
+Blocked: No for production alert-to-event packet deployment; yes for runtime remediation by design.
+Next: Future firing alert must produce AI triage packet and dry-run evidence first. `HostCiRunnerLoadSaturation` remains capacity / runner scheduling triage, not process kill. Runtime remediation remains `0` until owner approval, maintenance window, evidence ref, gated SIGTERM, post-check, and KM / PlayBook / Verifier writeback exist.
+Completion: host runaway monitoring / alert / PlayBook / Telegram event packet / production deploy readback 100%; runtime auto-remediation remains safely gated at 0.
+
 2026-06-18 13:43 Asia/Taipei
 Phase: P1/P2/P3 live readback
 Before: live cold-start was `PASS=83 WARN=1 BLOCKED=0`, result `DEGRADED`, because retained stale `km-vectorize-29689620` failed Job evidence was still counted as a service warning.