feat(api): Phase 7.5-7.6 Playbook 整合決策與自動萃取

Phase 7.5: DecisionManager 三軌決策
- 新增 Playbook 優先匹配 (similarity >= 85%)
- 三軌決策順序: Playbook > LLM > Expert System
- 整合 PlaybookService 推薦引擎

Phase 7.6: 自動萃取機制
- approval_execution.py 成功執行後觸發萃取
- 條件: RESOLVED/CLOSED + effectiveness >= 4
- 滿分 (5) 自動核准 Playbook

測試:
- 13 個 Playbook 單元測試全部通過
- 修復 Incident 模型欄位對應 (reasoning_steps)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

apps/api/src/services/decision_manager.py

@@ -30,10 +30,15 @@ import structlog
 from src.core.config import settings
 from src.core.redis_client import get_redis
 from src.models.incident import Incident
+from src.models.playbook import SymptomPattern
 from src.services.openclaw import get_openclaw
+from src.services.playbook_service import get_playbook_service
 
 logger = structlog.get_logger(__name__)
 
+# Phase 7.5: Playbook 優先閾值
+PLAYBOOK_SIMILARITY_THRESHOLD = 0.85  # 相似度 >= 85% 直接使用 Playbook
+
 
 # =============================================================================
 # Telegram 推送 (Phase 6.5: 決策就緒通知)
@@ -394,13 +399,20 @@ class DecisionManager:
         incident: Incident,
     ) -> dict[str, Any]:
         """
-        雙軌決策分析
+        三軌決策分析 (Phase 7.5 升級)
 
         策略:
-        - 同時啟動 LLM 和 Expert System
-        - LLM 成功則用 LLM (更智能)
-        - LLM 失敗則用 Expert System (保底)
+        1. 先檢查 Playbook 是否有高度匹配 (similarity >= 85%)
+        2. Playbook 命中則直接使用 (最快、經驗驗證)
+        3. 否則 LLM + Expert System 雙軌
+
+        優先順序: Playbook > LLM > Expert System
         """
+        # Phase 7.5: 先嘗試 Playbook 匹配
+        playbook_result = await self._try_playbook_match(incident)
+        if playbook_result:
+            return playbook_result
+
         # Expert System 同步執行 (立即可用)
         expert_result = expert_analyze(incident)
 
@@ -440,6 +452,108 @@ class DecisionManager:
         )
         return expert_result
 
+    async def _try_playbook_match(
+        self,
+        incident: Incident,
+    ) -> dict[str, Any] | None:
+        """
+        Phase 7.5: 嘗試 Playbook 匹配
+
+        條件:
+        - 相似度 >= PLAYBOOK_SIMILARITY_THRESHOLD (85%)
+        - Playbook 狀態為 APPROVED
+        - 成功率 >= 80% (如果有執行紀錄)
+
+        Returns:
+            匹配成功返回 proposal_data，否則 None
+        """
+        try:
+            playbook_service = get_playbook_service()
+
+            # 建構症狀模式
+            alert_names = [s.alert_name for s in incident.signals] if incident.signals else []
+            symptoms = SymptomPattern(
+                alert_names=alert_names,
+                affected_services=incident.affected_services or [],
+                severity_range=[incident.severity.value] if incident.severity else ["P2"],
+            )
+
+            # 取得推薦 (只取 Top 1)
+            recommendations = await playbook_service.get_recommendations(
+                symptoms=symptoms,
+                top_k=1,
+            )
+
+            if not recommendations:
+                logger.debug(
+                    "playbook_no_match",
+                    incident_id=incident.incident_id,
+                )
+                return None
+
+            best_match = recommendations[0]
+            playbook = best_match.playbook
+
+            # 檢查相似度閾值
+            if best_match.similarity_score < PLAYBOOK_SIMILARITY_THRESHOLD:
+                logger.debug(
+                    "playbook_similarity_below_threshold",
+                    incident_id=incident.incident_id,
+                    playbook_id=playbook.playbook_id,
+                    similarity=best_match.similarity_score,
+                    threshold=PLAYBOOK_SIMILARITY_THRESHOLD,
+                )
+                return None
+
+            # 檢查成功率 (如果有執行紀錄)
+            if playbook.total_executions > 0 and playbook.success_rate < 0.8:
+                logger.debug(
+                    "playbook_low_success_rate",
+                    incident_id=incident.incident_id,
+                    playbook_id=playbook.playbook_id,
+                    success_rate=playbook.success_rate,
+                )
+                return None
+
+            # Playbook 命中！
+            # 取得第一個修復步驟的指令
+            kubectl_command = ""
+            if playbook.repair_steps:
+                # 將 target 替換為實際服務名稱
+                target = incident.affected_services[0] if incident.affected_services else "unknown"
+                kubectl_command = playbook.repair_steps[0].command.format(target=target)
+
+            logger.info(
+                "playbook_match_success",
+                incident_id=incident.incident_id,
+                playbook_id=playbook.playbook_id,
+                playbook_name=playbook.name,
+                similarity=best_match.similarity_score,
+                success_rate=playbook.success_rate,
+            )
+
+            return {
+                "source": "playbook",
+                "playbook_id": playbook.playbook_id,
+                "playbook_name": playbook.name,
+                "action": kubectl_command,
+                "kubectl_command": kubectl_command,
+                "description": playbook.description,
+                "risk_level": playbook.repair_steps[0].risk_level.value.lower() if playbook.repair_steps else "medium",
+                "reasoning": f"Playbook 匹配 ({best_match.similarity_score:.0%} 相似度, {playbook.success_rate:.0%} 成功率): {best_match.reason}",
+                "confidence": min(best_match.similarity_score, playbook.success_rate) if playbook.total_executions > 0 else best_match.similarity_score,
+                "matched_symptoms": best_match.matched_symptoms,
+                "from_cache": False,
+            }
+
+        except Exception as e:
+            logger.warning(
+                "playbook_match_error",
+                incident_id=incident.incident_id,
+                error=str(e),
+            )
+            return None
+
     async def _find_existing_token(
         self,
         incident_id: str,