diff --git a/apps/api/src/agents/diagnostician_agent.py b/apps/api/src/agents/diagnostician_agent.py
index aed2f8b2..28990cd1 100644
--- a/apps/api/src/agents/diagnostician_agent.py
+++ b/apps/api/src/agents/diagnostician_agent.py
@@ -102,9 +102,21 @@ class DiagnosticianAgent(BaseAgent):
             "anomaly_context": snapshot.anomaly_context,
         })
 
+        # 2026-04-16 ogt + Claude Sonnet 4.6: 傳遞 snapshot 結構化資料給 OPENCLAW_NEMO
+        # 根因：原本 call(prompt) 不傳 context → nemo fallback 把 prompt[:500]（系統說明）
+        #       當 signal description → LLM 回傳 "調查 AWOOOI SRE 系統的偵探 Agent" 垃圾
+        # 修復：把 snapshot.evidence_summary 放進 alert_context.signals 讓 nemo 看到真實資料
+        _evidence = (snapshot.evidence_summary or "（待感應器資料）")[:800]
+        alert_context = {
+            "incident_id": snapshot.snapshot_id or "UNKNOWN",
+            "severity": "P3",
+            "signals": [{"alert_name": "evidence_snapshot", "description": _evidence}],
+            "affected_services": [],
+        }
+
         from src.services.openclaw import get_openclaw
         openclaw = get_openclaw()
-        response_text, _provider, success = await openclaw.call(prompt)
+        response_text, _provider, success = await openclaw.call(prompt, alert_context=alert_context)
 
         if not success or not response_text:
             return self._degraded_report(snapshot, 0, reason="llm_failed")
@@ -223,8 +235,13 @@ def _extract_hypotheses(parsed: dict[str, Any]) -> list[Hypothesis]:
                        "medium": "ModerateIssue", "low": "LowRisk"}
         category = risk_to_cat.get(risk_level.lower(), "Unknown")
         if action_title and confidence > 0:
+            # 2026-04-16 ogt + Claude Sonnet 4.6: 優先用 reasoning 作為假設描述
+            # reasoning（解釋「為什麼」採取行動）比 action_title（「做什麼」）更接近根因
+            # 例: reasoning="CPU 95%, 系統過載" vs action_title="重啟 Pod"
+            nemo_reasoning = str(parsed.get("reasoning", "")).strip()
+            description = nemo_reasoning[:500] if len(nemo_reasoning) > 20 else action_title[:500]
             return [Hypothesis(
-                description=action_title[:500],
+                description=description,
                 confidence=confidence,
                 evidence_chain=[],
                 category=category,
diff --git a/apps/api/src/services/openclaw.py b/apps/api/src/services/openclaw.py
index 0b01441a..46c30859 100644
--- a/apps/api/src/services/openclaw.py
+++ b/apps/api/src/services/openclaw.py
@@ -810,7 +810,11 @@ class OpenClawService:
     # Public LLM Interface (ILLMProvider Protocol)
     # =========================================================================
 
-    async def call(self, prompt: str) -> tuple[str, str, bool]:
+    async def call(
+        self,
+        prompt: str,
+        alert_context: dict | None = None,
+    ) -> tuple[str, str, bool]:
         """
         呼叫 LLM (ILLMProvider Protocol 實作)
 
@@ -818,14 +822,20 @@ class OpenClawService:
 
         Args:
             prompt: 完整的 prompt
+            alert_context: 可選的告警上下文（含 incident_id/signals，
+                供 OPENCLAW_NEMO provider 使用結構化 API）
 
         Returns:
             (response, provider, success)
 
         2026-04-16 ogt + Claude Sonnet 4.6: 修復 — _call_with_fallback 回傳 5 值，
             call() 回傳 3 值，避免呼叫端 (diagnostician_agent:107) ValueError
+        2026-04-16 ogt + Claude Sonnet 4.6: 加入 alert_context — 讓 diagnostician
+            能傳 snapshot 結構化資料給 OPENCLAW_NEMO，避免 prompt 被截斷為 garbage
         """
-        response, provider, success, _tokens, _cost = await self._call_with_fallback(prompt)
+        response, provider, success, _tokens, _cost = await self._call_with_fallback(
+            prompt, alert_context=alert_context
+        )
         return response, provider, success
 
     # =========================================================================