Tighten chart guards and EA envelopes

TODO_NEXT_STEPS.txt

@@ -4,6 +4,7 @@
 ================================================================================
 
 【已完成】
+   - V10.460 收斂 daily/growth 圖表空白誤判與 ElephantAlpha 告警信封：`page-daily-sales.js`、`page-growth.js` 的 chart 判斷改為至少有一個非零資料點才繪製 Chart.js，避免全 0 序列只畫座標軸；`resource_optimization` / `ea_escalation` 改輸出 deterministic `decision_envelope`，只使用 action_plans、CPU 實測與 hygiene evidence，不再輸出空泛「48 小時效益」敘事。
    - V10.459 強化 PChome `protected_existing_match` 決策封包：解析 `existing_match_conflict` 的既有候選、新候選與雙方 score，寫入 `decision_envelope.evidence` / `expected_impact` / `guardrails`，並把下一步明確標成「比較既有正式候選與新候選」；仍保持 `can_auto_execute=false`，避免新候選分數較高時繞過人工覆核自動覆蓋正式價差。
    - V10.458 將 OpenClaw / 競品 PPT 接上 PChome 覆核 `decision_envelope` 摘要：`competitor_intel_repository.summarize_review_decision_envelopes()` 成為共用 formatter，OpenClaw 週報/日報/月報與競品簡報 data_summary / KPI slide 都讀同一份信封文字，避免策略報告與 PPT 各自翻譯覆核狀態或遺失 HITL guardrails。
    - V10.457 將 PChome 覆核 `decision_envelope` 連到人工操作面：Dashboard 覆核卡新增決策等級、資料品質、HITL/trace 信封摘要；`/api/export/excel/pchome-review` 匯出同步增加決策信封 ID、決策類型、建議代碼、責任人、資料品質、自動執行允許與證據摘要，讓線上操作與下載檔都保留同一份 guardrails。

config.py

@@ -325,7 +325,7 @@ YOUTUBE_API_KEY = os.getenv('YOUTUBE_API_KEY', '')
 # ==========================================
 # 系統版本與路徑
 # ==========================================
-SYSTEM_VERSION = "V10.459"
+SYSTEM_VERSION = "V10.460"
 LOG_FILE_PATH = os.path.join(BASE_DIR, 'logs/system.log')
 public_url = PUBLIC_URL  # 用於模板顯示
 

docs/AI_INTELLIGENCE_MODULE_SOT.md

@@ -2,7 +2,7 @@
 
 > **最後更新**: 2026-05-24 (台北時間)
 > **狀態**: 🟢 四 AI Agent 自動化閉環已落地；LLM 路由紅線升級為 Ollama-first 三主機級聯，Gemini 備援預設關閉
-> **適用版本**: V10.459
+> **適用版本**: V10.460
 
 ---
 
@@ -50,6 +50,7 @@
 - PChome 覆核隊列本身也必須輸出 `decision_envelope`：`fetch_competitor_review_queue()`、`fetch_competitor_review_queue_page()` 與 `/api/pchome-review/queue` 的每筆候選需帶相同的 `subject`、`evidence`、`recommended_action`、`expected_impact` 與 `guardrails`，供 Dashboard、Agent、Telegram 與 PPT 共用；任何下游不得另寫一套比價狀態翻譯或繞過 HITL guardrails。
 - Dashboard 覆核卡與 `/api/export/excel/pchome-review` 也必須顯示/匯出 `decision_envelope` 的等級、資料品質、建議代碼、HITL、trace 與 `can_auto_execute=false` 邊界；操作員離開系統畫面或下載 Excel 後，仍要看得到「不可自動寫正式價差」的 guardrails。
 - OpenClaw 週報/日報/月報與 competitor PPT 不得再各自重算或翻譯 PChome 覆核狀態；必須透過 `competitor_intel_repository.summarize_review_decision_envelopes()` 讀取同一份 `decision_envelope` 摘要，並在 prompt / data_summary / KPI slide 保留 HITL 與 `can_auto_execute=false` 邊界。
+- ElephantAlpha 的 `resource_optimization` 與低信心 `ea_escalation` 也必須輸出 `decision_envelope`：資源壓力信封只能使用 `action_plans`、CPU 實測、hygiene 結果與 insight/action trace，不得加入 LLM 預測效益；`triaged_alert()` 對 `ea_escalation` 亦需渲染信封並以 `decision_id` 作為 callback 追蹤 ID。
 
 ## 一、四 AI Agent 路由架構
 
@@ -148,6 +149,7 @@ SQL漏斗(~300筆)
 - ElephantAlpha 價格類 trigger 的 HITL / 決策 prefetch 必須先使用觸發 SQL 與 `competitor_prices` / `price_records` 的 DB 實證生成 SKU、MOMO / PChome 價差與建議 action lines；完整 Hermes LLM prefetch 預設關閉（`ELEPHANT_ALPHA_HERMES_LLM_PREFETCH_ENABLED=false`），避免 5s timeout 後落入無實證摘要或雲端備援。若無 DB 實證，只記錄 suppressed telemetry / cooldown，不發 Telegram 空告警。
 - ElephantAlpha 協調器收到非純 JSON、fenced JSON 或混文字 JSON 時，必須先做容錯抽取；仍無法解析時，只能使用 DB/Hermes 實證生成保守 HITL fallback。fallback 不得放入 OpenClaw `generate_*` 類舊策略步驟，也不得暗示已自動調價。
 - `resource_optimization` 不再交給 LLM 生成「預期效益 / 已執行」敘事，顯示名稱統一為「資源壓力治理」。此 trigger 必須先由程式量測 `action_plans` backlog、P1/P2 數、pending_review、逾時項目與 CPU load；只有 CPU 達門檻、P1/P2 積壓或逾時積壓才發 Telegram「資源壓力告警」。單純 queue 大但 CPU 正常只記錄 telemetry，不派發 Hermes/NemoTron、不宣稱 48 小時效益；Telegram 段落使用「系統處置紀錄」而非泛稱「已執行」，避免暗示 AI 已完成未經驗證的外部動作。
+- `resource_optimization` 的 Telegram 必須包含 `decision_envelope` 區塊，標明 `source_agent=elephant_alpha`、資料品質、量測證據、`can_auto_execute=false` 與 deterministic trace；此路徑不呼叫 Gemini、不呼叫 Hermes/NemoTron，也不得把 queue backlog 翻譯成主機資源耗盡。
 - `resource_optimization` 會先執行 `ActionPlanHygieneService` 清理過期噪音：只關閉超過 72 小時的 `code_review_fix` / `openclaw_recommendation` 類 advisory action_plans，以及 NemoTron `direct_response/reply_simple` 舊聊天回覆計畫；將狀態改為 `auto_disabled` 或 `rejected` 並寫入 `metadata_json.hygiene_history`。不刪資料，也不碰 NemoTron human_review / pricing / tool action 類業務行動。
 - `momo-scheduler` 每 6 小時固定執行 `run_action_plan_hygiene_task()`，讓過期 advisory action_plans 的關閉不再依賴 `resource_optimization` 告警觸發；排程失敗會經 EventRouter 發送 `action_plan_hygiene_failure`。
 - `action_plans` 產生端必須防重：Code Review 同一檔案已有 active `code_review_fix` 時不重建；OpenClaw recommendation 會寫入文字 fingerprint 並跳過同一建議；AIOrchestrator 不再把 NemoTron `direct_response/reply_simple` 聊天回覆存成 action plan，真正需工具、審核或執行的 NemoTron action 才能進 queue。

docs/memory/current_execution_queue_20260524.md

@@ -49,6 +49,7 @@
 - 2026-05-24 23:15 CST 起，Dashboard 覆核卡與 PChome 覆核 Excel 匯出也顯示/輸出信封摘要、資料品質、HITL、trace、自動執行阻擋原因與證據摘要；下載檔不得丟失 guardrails。
 - 2026-05-24 23:25 CST 起，OpenClaw 週報/日報/月報與 competitor PPT 使用 `summarize_review_decision_envelopes()` 的同一份 HITL 信封摘要，不再手寫 attempt_status 統計或自行翻譯覆核狀態。
 - 2026-05-24 23:40 CST 起，`compare_existing_identity` 成為 `protected_existing_match` 的明確建議動作；Agent 只能提示「比較既有正式候選與新候選」，不得因新候選分數較高自動寫正式價差。
+- 2026-05-24 23:55 CST 起，ElephantAlpha `resource_optimization` / `ea_escalation` 都必須帶 deterministic `decision_envelope`；Telegram 按鈕 callback 使用 `decision_id`，證據只允許 action_plans、CPU 實測、hygiene 與 trigger trace。
 - 告警不得再輸出空泛「預期效益」；必須帶資料品質、證據來源、HITL 邊界與 trace id。
 - Agent 建議只能輔助排序與分析，不得繞過 matcher / feeder / review service 寫正式價格。
 
@@ -58,6 +59,7 @@
 - `/daily_sales`、`/growth_analysis` 圖表不得空白；需保留原本圖表並升級成更專業的呈現。
 - 圖表需通過 runtime nonblank canvas 檢查與手機版 responsive。
 - daily/growth/PPT 必須共用 `competitor_intel_repository` 的比價資料出口，避免價差方向或統計口徑分裂。
+- 2026-05-24 23:55 CST 起，daily/growth chart 判斷不再只看 series 長度；若序列全 0，顯示 chart-empty 狀態而不是畫只有座標軸的假圖。正式 smoke 需跑 `scripts/check_sales_charts_runtime.js` 確認主要 canvas 非空。
 
 ## 5. PPT 視覺 QA 與自動簡報產線
 

docs/memory/history_logs.md

@@ -13,6 +13,7 @@
 ## 📅 詳細更新日誌 (考古存檔)
 
 ### 2026-05-24：PChome 近門檻身份回收第二輪
+- **V10.460 ElephantAlpha 告警決策信封**: `resource_optimization` 會為資源壓力告警產生 deterministic `decision_envelope`，證據只來自 `action_plans`、CPU 實測與 hygiene 結果，Telegram 同時顯示決策信封、量測指標、判讀、系統處置與下一步；`ea_escalation` 模板也會渲染信封並使用 `decision_id` 作為 `momo:eig:*` callback，避免低信心升級告警只剩空泛文字或不可追蹤按鈕。
 - **V10.459 protected_existing_match 決策封包**: PChome 覆核信封開始解析 `existing_match_conflict`，把既有正式候選、新候選、雙方 matcher score 與 score delta 寫入 evidence / expected_impact / guardrails；新候選即使分數較高也維持 `can_auto_execute=false`，但 OpenClaw、PPT、Dashboard 與人工覆核可清楚看見該比較哪兩個候選。
 - **V10.458 OpenClaw / PPT 決策信封摘要**: 新增 `summarize_review_decision_envelopes()` 作為 PChome 覆核信封共用摘要 formatter；OpenClaw 週報/日報/月報、OpenClaw Bot competitor PPT data_summary 與 PPT KPI slide 都使用同一份 HITL / 資料品質 / action / trace 摘要，不再各自手寫 attempt_status 翻譯。
 - **V10.457 Dashboard / Excel 決策信封連動**: 商品看板 PChome 覆核卡顯示 `decision_envelope` 的決策等級、資料品質、HITL 與 trace；`/api/export/excel/pchome-review` 匯出新增決策信封 ID、建議代碼、責任人、資料品質、自動執行允許、阻擋原因與證據摘要，讓下載檔仍保留不可自動寫正式價差的 guardrails。

services/elephant_alpha_autonomous_engine.py

@@ -1107,12 +1107,114 @@ class ElephantAlphaAutonomousEngine:
         actions.append("確認 action_plans 來源是否持續產生重複建議；若是報表型建議，應改為摘要消化而非逐筆告警。")
         return actions
 
+    @staticmethod
+    def _resource_pressure_severity(metrics: Dict[str, Any]) -> str:
+        level = str((metrics or {}).get("pressure_level") or "unknown")
+        if level == "critical":
+            return "P1"
+        if level == "warning":
+            return "P2"
+        if level == "backlog_only":
+            return "P3"
+        return "P4"
+
+    @staticmethod
+    def _build_resource_pressure_decision_envelope(
+        metrics: Dict[str, Any],
+        insight_id: Optional[int],
+        previous_limit: int,
+        new_limit: int,
+    ) -> Dict[str, Any]:
+        metrics = dict(metrics or {})
+        load_pct = float(metrics.get("system_load_pct") or 0.0)
+        hygiene = metrics.get("hygiene_result") if isinstance(metrics.get("hygiene_result"), dict) else {}
+        hygiene_count = int((hygiene or {}).get("updated_count") or 0)
+        data_quality = "partial" if metrics.get("query_error") else "complete"
+        deadline = (datetime.now() + timedelta(hours=2)).strftime("%Y-%m-%d %H:%M:%S")
+
+        evidence = [
+            {
+                "type": "queue",
+                "metric": "action_queue_size",
+                "value": f"{int(metrics.get('action_queue_size') or 0)}/{int(metrics.get('queue_threshold') or RESOURCE_QUEUE_THRESHOLD)}",
+                "basis": "action_plans pending / auto_pending / pending_review",
+                "confidence": 1.0 if data_quality == "complete" else 0.7,
+            },
+            {
+                "type": "priority",
+                "metric": "p1_p2_pending",
+                "value": f"{int(metrics.get('high_priority_count') or 0)}/{int(metrics.get('high_priority_threshold') or RESOURCE_HIGH_PRIORITY_THRESHOLD)}",
+                "basis": "priority <= 2",
+                "confidence": 1.0 if data_quality == "complete" else 0.7,
+            },
+            {
+                "type": "resource",
+                "metric": "cpu_load_pct",
+                "value": f"{load_pct:.1f}%/{float(metrics.get('load_threshold_pct') or RESOURCE_LOAD_THRESHOLD_PCT):.0f}%",
+                "basis": "local system load measurement",
+                "confidence": 0.9,
+            },
+        ]
+        if int(metrics.get("stale_count") or 0) > 0:
+            evidence.append({
+                "type": "stale",
+                "metric": "stale_action_plans",
+                "value": f"{int(metrics.get('stale_count') or 0)}/{int(metrics.get('stale_threshold') or RESOURCE_STALE_THRESHOLD)}",
+                "basis": f"age >= {int(metrics.get('stale_hours') or RESOURCE_STALE_HOURS)}h",
+                "confidence": 1.0 if data_quality == "complete" else 0.7,
+            })
+
+        return {
+            "decision_id": f"ea_resource_pressure_{insight_id or int(datetime.now().timestamp())}",
+            "decision_type": "resource_optimization",
+            "source_agent": "elephant_alpha",
+            "severity": ElephantAlphaAutonomousEngine._resource_pressure_severity(metrics),
+            "confidence": float(metrics.get("confidence") or 0.86),
+            "analysis": (
+                "由 action_plans 與 CPU 實測值判定資源壓力；"
+                "未使用 LLM 生成效益預測，也未啟動價格分析。"
+            ),
+            "subject": {
+                "sku": "action_plans",
+                "name": "Elephant Alpha 資源壓力治理",
+            },
+            "evidence": evidence,
+            "recommended_action": {
+                "action": "human_review_backlog_triage",
+                "owner": "ops",
+                "deadline": deadline,
+                "requires_hitl": True,
+            },
+            "expected_impact": {
+                "risk_reduction": (
+                    "reduce stale high-priority backlog"
+                    if metrics.get("should_alert") or hygiene_count > 0
+                    else "observe only"
+                ),
+            },
+            "guardrails": {
+                "can_auto_execute": False,
+                "blocked_reason": (
+                    "resource_optimization 只允許清理過期 advisory action_plans；"
+                    "外部修復、價格分析與策略派發需人工覆核"
+                ),
+                "data_quality": data_quality,
+                "llm_used": False,
+            },
+            "trace": {
+                "insight_id": insight_id,
+                "model": "deterministic_metrics",
+                "provider": "action_plans_cpu_probe",
+            },
+        }
+
     @staticmethod
     def _build_resource_pressure_telegram_message(
         metrics: Dict[str, Any],
         insight_id: Optional[int],
         previous_limit: int,
         new_limit: int,
+        decision_envelope: Optional[Dict[str, Any]] = None,
     ) -> str:
         level = str(metrics.get("pressure_level", "unknown"))
         level_label = {
@@ -1156,6 +1258,13 @@ class ElephantAlphaAutonomousEngine:
             "",
             "<b>量測指標</b>",
         ]
+        if decision_envelope:
+            try:
+                from services.telegram_templates import _format_decision_envelope
+
+                lines[4:4] = ["", *_format_decision_envelope(decision_envelope)]
+            except Exception:
+                pass
         if pre_hygiene:
             lines.append(
                 "• 清理前 Action queue："
@@ -1217,6 +1326,12 @@ class ElephantAlphaAutonomousEngine:
                 insight_id,
                 previous_limit,
                 new_limit,
+                decision_envelope=self._build_resource_pressure_decision_envelope(
+                    metrics,
+                    insight_id,
+                    previous_limit,
+                    new_limit,
+                ),
             )
             await self._run_with_timeout(_send_telegram_raw, msg, timeout=10)
             self._log.info("Resource pressure Telegram sent: level=%s", metrics.get("pressure_level"))
@@ -1699,6 +1814,72 @@ class ElephantAlphaAutonomousEngine:
                 exc_info=True,
             )
 
+    @staticmethod
+    def _build_human_escalation_decision_envelope(
+        decision: StrategicDecision,
+        trigger: AutonomousTrigger,
+        *,
+        insight_id: Optional[int] = None,
+        concrete_actions: Optional[List[str]] = None,
+    ) -> Dict[str, Any]:
+        concrete_actions = [str(item).strip() for item in (concrete_actions or []) if str(item).strip()]
+        data_quality = "complete" if concrete_actions else "partial"
+        trigger_label = _zh_trigger(trigger.trigger_type)
+        evidence = [
+            {
+                "type": "confidence",
+                "metric": "decision_confidence",
+                "value": f"{float(decision.confidence or 0):.2f}",
+                "basis": "ElephantAlpha confidence below autonomous threshold",
+                "confidence": float(decision.confidence or 0),
+            },
+            {
+                "type": "trigger",
+                "metric": "trigger_type",
+                "value": trigger.trigger_type,
+                "basis": trigger_label,
+            },
+        ]
+        for idx, action in enumerate(concrete_actions[:3], start=1):
+            evidence.append({
+                "type": "action_candidate",
+                "metric": f"candidate_{idx}",
+                "value": action[:120],
+                "basis": "DB/Hermes concrete evidence" if trigger.trigger_type in _PRICE_RELATED_TRIGGERS else "trigger evidence",
+            })
+
+        return {
+            "decision_id": f"ea_review_{insight_id or int(datetime.now().timestamp())}",
+            "decision_type": "ea_escalation",
+            "source_agent": "elephant_alpha",
+            "severity": "P2" if data_quality == "complete" else "P3",
+            "confidence": float(decision.confidence or 0),
+            "analysis": "低信心自主決策已轉人工覆核；未批准前不執行外部副作用。",
+            "subject": {
+                "sku": trigger.trigger_type,
+                "name": f"Elephant Alpha · {trigger_label}",
+            },
+            "evidence": evidence,
+            "recommended_action": {
+                "action": "human_review",
+                "owner": "ops",
+                "requires_hitl": True,
+            },
+            "expected_impact": {
+                "risk_reduction": "prevent low-confidence autonomous execution",
+            },
+            "guardrails": {
+                "can_auto_execute": False,
+                "blocked_reason": "L3 HITL required; no automatic execution before approval",
+                "data_quality": data_quality,
+            },
+            "trace": {
+                "insight_id": insight_id,
+                "model": "deterministic_escalation_gate",
+                "provider": "elephant_alpha",
+            },
+        }
+
     async def _escalate_to_human(self, decision: StrategicDecision, trigger: AutonomousTrigger) -> None:
         self._log.warning("Escalating to human: %s", trigger.trigger_type)
         concrete_actions = (
@@ -1717,6 +1898,7 @@ class ElephantAlphaAutonomousEngine:
             self._record_suppressed_escalation(decision, trigger, "no_concrete_evidence")
             return
 
+        insight_id = None
         session = get_session()
         try:
             row = session.execute(
@@ -1744,6 +1926,7 @@ class ElephantAlphaAutonomousEngine:
             ).fetchone()
             session.commit()
             if row:
+                insight_id = row[0]
                 try:
                     from services.openclaw_learning_service import enqueue_insight_embedding
                     enqueue_insight_embedding(
@@ -1803,6 +1986,12 @@ class ElephantAlphaAutonomousEngine:
                 )
 
             try:
+                decision_envelope = self._build_human_escalation_decision_envelope(
+                    decision,
+                    trigger,
+                    insight_id=insight_id,
+                    concrete_actions=concrete_actions,
+                )
                 msg, keyboard = triaged_alert(
                     base_event={
                         "event_type": "ea_escalation",
@@ -1811,7 +2000,8 @@ class ElephantAlphaAutonomousEngine:
                             f"自主決策信心度 {decision.confidence:.2f} 低於門檻，需人工批准"
                             + ("" if concrete_actions else "（⚠️ 無實證數據）")
                         ),
-                        "id": f"ea_review_{int(datetime.now().timestamp())}",
+                        "id": decision_envelope.get("decision_id"),
+                        "decision_envelope": decision_envelope,
                     },
                     tier_label="🐘 Elephant Alpha · L3 HITL",
                     ai_summary=ai_summary_text,

services/telegram_templates.py

@@ -648,6 +648,7 @@ def _format_ea_escalation_alert(
     generic_actions = [item for item in parsed_actions if not _is_ea_sku_action(item)]
     shown_actions = sku_actions[:5]
     hidden_count = max(0, len(sku_actions) - len(shown_actions))
+    decision_envelope = base_event.get("decision_envelope") or base_event.get("decision")
 
     lines = [
         f"⚡ <b>{escape(str(tier_label))}</b>",
@@ -661,6 +662,9 @@ def _format_ea_escalation_alert(
     for part in cause_parts[:3]:
         lines.append(f"• {part}")
 
+    if isinstance(decision_envelope, dict) and decision_envelope:
+        lines += ["", *_format_decision_envelope(decision_envelope)]
+
     if ai_summary:
         lines += [
             "",
@@ -824,6 +828,9 @@ def triaged_alert(base_event: Dict[str, Any], tier_label: str,
     title = escape(str(base_event.get("title", "")))
     summary = escape(str(base_event.get("summary", "")))
     event_id = base_event.get("id")
+    decision_envelope = base_event.get("decision_envelope") or base_event.get("decision")
+    if not event_id and isinstance(decision_envelope, dict):
+        event_id = decision_envelope.get("decision_id")
     safe_ai_summary = escape(str(ai_summary or ""))
     safe_ai_cause = escape(str(ai_cause or "")) if ai_cause else None
     safe_actions = [escape(str(a)) for a in (ai_actions or [])]
@@ -849,7 +856,6 @@ def triaged_alert(base_event: Dict[str, Any], tier_label: str,
             lines += [f"🧠 <b>AI 摘要：</b>{safe_ai_summary[:400]}", ""]
         if safe_ai_cause:
             lines += [f"💡 <b>可能原因：</b>{safe_ai_cause}", ""]
-        decision_envelope = base_event.get("decision_envelope") or base_event.get("decision")
         if isinstance(decision_envelope, dict):
             lines += _format_decision_envelope(decision_envelope)
             if not event_id:

tests/test_chart_fallback_contract.py

@@ -14,6 +14,7 @@ def test_daily_sales_canvas_is_primary_and_fallback_is_opt_in():
     assert ".chart-container:not(.chart-fallback-active) .chart-fallback-list" in css
     assert "node.content && node.content.textContent" in script
     assert "chart-empty-active" in script
+    assert "Math.abs(Number(value)) > 1e-9" in script
     render_body = script.split("function renderAllCharts()", 1)[1].split("function bootCharts()", 1)[0]
     assert "renderHtmlChartFallbacks();" not in render_body
     assert "catch(error =>" in script
@@ -30,6 +31,7 @@ def test_growth_analysis_canvas_is_primary_and_fallback_is_opt_in():
     assert ".ga-chart-card__body:not(.chart-fallback-active) .ga-chart-snapshot" in css
     assert "node.content && node.content.textContent" in script
     assert "chart-empty-active" in script
+    assert "Math.abs(Number(value)) > 1e-9" in script
     render_body = script.split("function renderCharts()", 1)[1].split("function bootCharts()", 1)[0]
     assert "renderHtmlChartFallbacks();" not in render_body
     assert "catch(error =>" in script

tests/test_elephant_alpha_engine.py

@@ -373,6 +373,49 @@ def test_resource_pressure_message_is_measurement_based_not_llm_theatre():
     assert "48 小時效益預測" in msg
 
 
+def test_resource_pressure_decision_envelope_is_measurement_based():
+    from services.elephant_alpha_autonomous_engine import ElephantAlphaAutonomousEngine
+
+    metrics = ElephantAlphaAutonomousEngine._classify_resource_pressure({
+        "action_queue_size": 34,
+        "high_priority_count": 8,
+        "human_review_count": 6,
+        "stale_count": 5,
+        "system_load_pct": 22.0,
+        "queue_threshold": 10,
+        "load_threshold_pct": 80,
+        "high_priority_threshold": 5,
+        "stale_threshold": 5,
+        "stale_hours": 24,
+    })
+
+    envelope = ElephantAlphaAutonomousEngine._build_resource_pressure_decision_envelope(
+        metrics,
+        insight_id=1942,
+        previous_limit=10,
+        new_limit=8,
+    )
+    msg = ElephantAlphaAutonomousEngine._build_resource_pressure_telegram_message(
+        metrics,
+        insight_id=1942,
+        previous_limit=10,
+        new_limit=8,
+        decision_envelope=envelope,
+    )
+
+    assert envelope["decision_id"] == "ea_resource_pressure_1942"
+    assert envelope["source_agent"] == "elephant_alpha"
+    assert envelope["decision_type"] == "resource_optimization"
+    assert envelope["severity"] == "P2"
+    assert envelope["guardrails"]["can_auto_execute"] is False
+    assert envelope["guardrails"]["llm_used"] is False
+    assert envelope["trace"]["provider"] == "action_plans_cpu_probe"
+    assert "🧭 <b>決策信封</b>" in msg
+    assert "資料品質：<code>complete</code>　自動執行：<b>不允許</b>" in msg
+    assert "不採用 LLM 生成的 48 小時效益預測" in msg
+    assert "Gemini" not in msg
+
+
 def test_resource_optimization_bypasses_llm_orchestrator(monkeypatch):
     import services.elephant_alpha_autonomous_engine as engine_module
     from services.elephant_alpha_autonomous_engine import (
@@ -495,6 +538,44 @@ def test_resource_optimization_cannot_use_legacy_autonomous_execution_template(m
     assert sent == []
 
 
+def test_human_escalation_decision_envelope_blocks_auto_execution():
+    from services.elephant_alpha_autonomous_engine import (
+        AutonomousTrigger,
+        ElephantAlphaAutonomousEngine,
+    )
+    from services.elephant_alpha_orchestrator import StrategicDecision
+
+    decision = StrategicDecision(
+        priority="medium",
+        agents_required=["elephant_alpha"],
+        reasoning="需要人工判讀。",
+        expected_outcome="不自動執行。",
+        confidence=0.62,
+        execution_plan=[],
+        resource_requirements={},
+    )
+    trigger = AutonomousTrigger(
+        trigger_type="code_exception",
+        conditions={"scan_containers": ["momo-pro-system"]},
+        threshold=1.0,
+        enabled=True,
+    )
+
+    envelope = ElephantAlphaAutonomousEngine._build_human_escalation_decision_envelope(
+        decision,
+        trigger,
+        insight_id=77,
+        concrete_actions=[],
+    )
+
+    assert envelope["decision_id"] == "ea_review_77"
+    assert envelope["decision_type"] == "ea_escalation"
+    assert envelope["guardrails"]["can_auto_execute"] is False
+    assert envelope["guardrails"]["data_quality"] == "partial"
+    assert envelope["recommended_action"]["requires_hitl"] is True
+    assert envelope["trace"]["provider"] == "elephant_alpha"
+
+
 def test_elephant_alpha_openclaw_registry_is_ollama_first():
     from services.elephant_alpha_orchestrator import ElephantAlphaOrchestrator
 

tests/test_telegram_triaged_alert_format.py

@@ -126,6 +126,57 @@ def test_ea_escalation_generic_actions_do_not_render_as_sku_cards():
     assert "未取得實證前，不執行自動調價、修復或策略派發" in msg
 
 
+def test_ea_escalation_renders_decision_envelope_and_uses_decision_id_callback():
+    msg, keyboard = triaged_alert(
+        base_event={
+            "event_type": "ea_escalation",
+            "title": "🐘 EA 升級審核 · 資源壓力治理",
+            "summary": "自主決策信心度 0.82 低於門檻，需人工批准",
+            "decision_envelope": {
+                "decision_id": "ea_resource_pressure_1942",
+                "decision_type": "resource_optimization",
+                "source_agent": "elephant_alpha",
+                "severity": "P2",
+                "confidence": 0.86,
+                "subject": {
+                    "sku": "action_plans",
+                    "name": "Elephant Alpha 資源壓力治理",
+                },
+                "evidence": [
+                    {
+                        "type": "queue",
+                        "metric": "action_queue_size",
+                        "value": "7/10",
+                        "basis": "action_plans pending / pending_review",
+                        "confidence": 1.0,
+                    },
+                ],
+                "recommended_action": {
+                    "action": "human_review_backlog_triage",
+                    "owner": "ops",
+                    "requires_hitl": True,
+                },
+                "guardrails": {
+                    "can_auto_execute": False,
+                    "blocked_reason": "resource_optimization requires HITL",
+                    "data_quality": "complete",
+                },
+            },
+        },
+        tier_label="🐘 Elephant Alpha · L3 HITL",
+        ai_summary="只採用 action_plans 與 CPU 實測值。",
+        ai_cause="觸發類型：資源壓力治理 | 信心度：0.86",
+        ai_actions=["檢查 priority <= 2 的 action_plans"],
+    )
+
+    assert "🧭 <b>決策信封</b>" in msg
+    assert "類型：<code>resource_optimization</code>" in msg
+    assert "資料品質：<code>complete</code>　自動執行：<b>不允許</b>" in msg
+    assert "SKU：<code>action_plans</code>" in msg
+    assert "human_review_backlog_triage" in msg
+    assert keyboard["inline_keyboard"][0][0]["callback_data"] == "momo:eig:ea_resource_pressure_1942"
+
+
 def test_triaged_alert_renders_decision_envelope_contract():
     msg, keyboard = triaged_alert(
         base_event={

web/static/js/page-daily-sales.js

@@ -219,7 +219,10 @@
   function hasSeriesData(labels, ...seriesList) {
     return Array.isArray(labels) &&
       labels.length > 0 &&
-      seriesList.some(series => Array.isArray(series) && series.length > 0);
+      seriesList.some(series =>
+        Array.isArray(series) &&
+        series.some(value => Number.isFinite(Number(value)) && Math.abs(Number(value)) > 1e-9)
+      );
   }
 
   function renderChartEmpty(canvasId, message) {

web/static/js/page-growth.js

@@ -135,7 +135,10 @@
   function hasSeriesData(labels, ...seriesList) {
     return Array.isArray(labels) &&
       labels.length > 0 &&
-      seriesList.some(series => Array.isArray(series) && series.length > 0);
+      seriesList.some(series =>
+        Array.isArray(series) &&
+        series.some(value => Number.isFinite(Number(value)) && Math.abs(Number(value)) > 1e-9)
+      );
   }
 
   function renderChartEmpty(canvasId, message) {