feat(intelligence): Phase 5 主动行为激活 — 注入格式 + 跨会话连续性 + 触发持久化

Task 5.1+5.4: ButlerRouter/experience 注入格式升级为 <butler-context> XML fencing
- butler_router: [路由上下文] → <butler-context><routing>...</routing></butler-context>
- experience: [过往经验] → <butler-context><experience>...</experience></butler-context>
- 统一 system-note 提示，引导 LLM 自然运用上下文

Task 5.2: 跨会话连续性 — pre_conversation_hook 注入活跃痛点 + 相关经验
- 从 VikingStorage 检索相关记忆(相似度>=0.3)
- 从 pain_aggregator 获取 High severity 痛点(top 3)

Task 5.3: 触发信号持久化 — post_conversation_hook 将触发信号存入 VikingStorage
- store_trigger_experience(): 模板提取，零 LLM 成本
- 为未来 LLM 深度反思积累数据基础

crates/zclaw-runtime/src/middleware/butler_router.rs

@@ -201,12 +201,15 @@ impl ButlerRouterMiddleware {
     }
 
     /// Domain context to inject into system prompt based on routing hint.
+    ///
+    /// Uses structured `<butler-context>` XML fencing (Hermes-inspired) for
+    /// reliable prompt cache preservation across turns.
     fn build_context_injection(hint: &RoutingHint) -> String {
         // Semantic skill routing
         if hint.category == "semantic_skill" {
             if let Some(ref skill_id) = hint.skill_id {
                 return format!(
-                    "\n\n[语义路由] 匹配技能: {} (置信度: {:.0}%)\n系统检测到用户的意图与已注册技能高度相关，请在回答中充分利用该技能的能力。",
+                    "\n\n<butler-context>\n<routing>匹配技能: {} (置信度: {:.0}%)</routing>\n<system-note>系统检测到用户的意图与已注册技能高度相关，请在回答中充分利用该技能的能力。</system-note>\n</butler-context>",
                     skill_id,
                     hint.confidence * 100.0
                 );
@@ -230,11 +233,11 @@ impl ButlerRouterMiddleware {
         }
 
         let skill_info = hint.skill_id.as_ref().map_or(String::new(), |id| {
-            format!("\n关联技能: {}", id)
+            format!("\n<skill>{}</skill>", id)
         });
 
         format!(
-            "\n\n[路由上下文] (置信度: {:.0}%)\n{}{}",
+            "\n\n<butler-context>\n<routing confidence=\"{:.0}%\">{}</routing>{}<system-note>以上是管家系统对您当前意图的分析。在对话中自然运用这些信息，主动提供有帮助的建议。</system-note>\n</butler-context>",
             hint.confidence * 100.0,
             domain_context,
             skill_info
@@ -357,7 +360,7 @@ mod tests {
             domain_prompt: None,
         };
         let injection = ButlerRouterMiddleware::build_context_injection(&hint);
-        assert!(injection.contains("路由上下文"));
+        assert!(injection.contains("butler-context"));
         assert!(injection.contains("医院"));
         assert!(injection.contains("80%"));
     }
@@ -435,7 +438,7 @@ mod tests {
 
         let decision = mw.before_completion(&mut ctx).await.unwrap();
         assert!(matches!(decision, MiddlewareDecision::Continue));
-        assert!(ctx.system_prompt.contains("路由上下文"));
+        assert!(ctx.system_prompt.contains("butler-context"));
         assert!(ctx.system_prompt.contains("医院"));
     }
 
@@ -464,7 +467,7 @@ mod tests {
 
         let decision = mw.before_completion(&mut ctx).await.unwrap();
         assert!(matches!(decision, MiddlewareDecision::Continue));
-        assert!(ctx.system_prompt.contains("路由上下文"));
+        assert!(ctx.system_prompt.contains("butler-context"));
         assert!(ctx.system_prompt.contains("电商运营管家"));
     }