diff --git a/.agents/skills/02-lewooogo-backend-core.md b/.agents/skills/02-lewooogo-backend-core.md
index 79799864..a8e22510 100644
--- a/.agents/skills/02-lewooogo-backend-core.md
+++ b/.agents/skills/02-lewooogo-backend-core.md
@@ -10,10 +10,10 @@
 
 | 欄位 | 值 |
 |------|-----|
-| **版本** | v2.1 |
+| **版本** | v2.2 |
 | **建立日期** | 2026-03-20 (台北) |
 | **建立者** | Claude Code |
-| **最後修改** | 2026-03-29 20:10 (台北) |
+| **最後修改** | 2026-03-30 03:30 (台北) |
 | **修改者** | Claude Code |
 
 ### 變更紀錄
@@ -32,6 +32,7 @@
 | v1.9 | 2026-03-28 | Claude Code | 🦞 新增 Phase 19 Terminal SSE 後端整合章節 |
 | v2.0 | 2026-03-29 | Claude Code | 🔴 Phase D-G P0 修正: 新增 LearningRepository (積木化合規) |
 | v2.1 | 2026-03-29 | Claude Code | 🛡️ 新增 ADR-038/039 OpenClaw 安全網章節 (Wave 1 部署) |
+| v2.2 | 2026-03-30 | Claude Code | 🎯 新增 Phase 19.4 API 整合章節 (Approval + K8s 真實 API) |
 
 ---
 
@@ -803,6 +804,58 @@ class SSEEventType(str, Enum):
 | 重連機制 | 指數退避 + Last-Event-ID 續接 |
 | Ghost Payload | 最小化：只傳 `current_page` + `entity_id` |
 
+### Phase 19.4 API 整合模式 (2026-03-30)
+
+> **首席架構師審查**: 94/100 OUTSTANDING
+> **測試覆蓋**: 60 測試全部通過
+
+#### 正確做法 - 透過 DI 取得 Service
+
+```python
+# ✅ 正確: 透過 DI 取得 ApprovalDBService
+from src.services.approval_db import get_approval_service
+
+async def _handle_approval_action(self, publisher, topic, ...):
+    approval_service = get_approval_service()
+    pending_approvals = await approval_service.get_pending_approvals()
+
+    # 渲染 ApprovalCard
+    await self._publish_render_ui(publisher, topic, "ApprovalCard", {
+        "approvalId": str(approval.id)[:8].upper(),
+        "riskLevel": approval.risk_level.value.upper(),
+        # ...
+    })
+```
+
+#### K8s 狀態查詢
+
+```python
+# ✅ 正確: 處理 K8s 不可用情況
+from src.services.k8s_diagnostics import _get_k8s_client
+
+async def _handle_status_query(self, publisher, topic, ...):
+    client = await _get_k8s_client()
+    if not client:
+        await self._publish_thought(publisher, topic, "System", "K8s 連線失敗")
+        return
+
+    v1 = client.CoreV1Api()
+    pods = await v1.list_namespaced_pod(namespace="awoooi")
+```
+
+#### 錯誤處理鐵律
+
+```python
+# ✅ 正確: 完整錯誤處理 + structlog
+try:
+    result = await service.some_operation()
+except Exception as e:
+    logger.error("operation_failed", error=str(e))
+    await self._publish_tool_call(publisher, topic, "Tool", "failed",
+        result={"error": str(e)[:100]}  # 截斷避免敏感資訊
+    )
+```
+
 ---
 
 ## 參考文檔
diff --git a/docs/LOGBOOK.md b/docs/LOGBOOK.md
index ed56c381..9a9abe84 100644
--- a/docs/LOGBOOK.md
+++ b/docs/LOGBOOK.md
@@ -5,14 +5,14 @@
 
 ---
 
-## 📍 當前狀態 (2026-03-30 03:00 台北)
+## 📍 當前狀態 (2026-03-30 03:30 台北)
 
 | 項目 | 狀態 |
 |------|------|
 | **AI 仲裁** | ✅ **NVIDIA 優先** `["nvidia","gemini","ollama","claude"]` |
 | **CI/CD 告警** | ✅ **簡化格式** (跳過 AI 仲裁) |
 | **P0 安全修復** | ✅ **sudo 密碼明文移除** (1cec655) |
-| **首席架構師** | ✅ **78→85/100 通過** |
+| **首席架構師** | ✅ **94/100 OUTSTANDING** (Phase 19.4 + ADR-039 + AI 仲裁) |
 | **CD 佇列模式** | ✅ **cancel-in-progress: false** |
 | **Phase 19.4** | ✅ **Terminal Service API 整合完成** (60 測試通過) |
 | **Intent Classifier** | ✅ **Ollama 整合完成** (21 測試通過) |
diff --git a/docs/adr/ADR-030-intelligent-auto-remediation.md b/docs/adr/ADR-030-intelligent-auto-remediation.md
index 29a166c5..1953e0ae 100644
--- a/docs/adr/ADR-030-intelligent-auto-remediation.md
+++ b/docs/adr/ADR-030-intelligent-auto-remediation.md
@@ -632,6 +632,27 @@ class TrustEngine:
 - 修改 `apps/api/src/services/trust_engine.py`
 - 修改 `apps/api/src/services/approval_execution.py`
 
+#### 5.3 Playbook 自動狀態轉換 (2026-03-30 補充)
+
+> **實作位置**: `apps/api/src/repositories/playbook_repository.py:adjust_confidence()`
+
+| 狀態轉換 | 觸發條件 | 說明 |
+|---------|---------|------|
+| **DRAFT → APPROVED** | `confidence >= 0.9` | 高信心度自動升級 |
+| **任意 → DEPRECATED** | `confidence < 0.3` + `failure_rate > 50%` + `executions >= 5` | 低效 Playbook 自動棄用 |
+
+```python
+# Playbook 信心度調整常數
+CONFIDENCE_PROMOTE_THRESHOLD = 0.9   # 自動升級閾值
+CONFIDENCE_DEPRECATE_THRESHOLD = 0.3 # 自動棄用閾值
+FAILURE_RATE_THRESHOLD = 0.5         # 失敗率閾值
+MIN_EXECUTIONS_FOR_DEPRECATE = 5     # 最小執行次數
+
+# Learning Service 信心度調整
+PROMOTE_DELTA = +0.1   # 高評分 (>=4) +10%
+DEMOTE_DELTA = -0.15   # 低評分 (<=2) -15%
+```
+
 ---
 
 ## 四、架構相容性分析