diff --git a/apps/web/src/app/[locale]/page.tsx b/apps/web/src/app/[locale]/page.tsx
index 9c7ee0fbd..e5f5cd3b4 100644
--- a/apps/web/src/app/[locale]/page.tsx
+++ b/apps/web/src/app/[locale]/page.tsx
@@ -31,12 +31,95 @@ import {
   IncidentCardGrid,
   IncidentEmptyState,
   ThinkingTerminal,
-  DEMO_DECISION_CHAIN,
   DualStateIncidentCard,
 } from '@/components/incident'
 import { AlertTriangle } from 'lucide-react'
 import type { IncidentResponse } from '@/lib/api-client'
 
+// =============================================================================
+// Types for AI Decision Chain Display
+// =============================================================================
+
+interface ReasoningStep {
+  step: string
+  reasoning: string
+  timestamp?: string
+}
+
+interface DecisionChainDisplay {
+  analysis_type: 'blast_radius' | 'root_cause' | 'action_suggestion'
+  target_service: string
+  reasoning_steps: ReasoningStep[]
+  conclusion: string
+  confidence: number
+}
+
+// =============================================================================
+// Utility: Convert API proposal_data to DecisionChain format
+// =============================================================================
+
+function convertToDecisionChain(
+  incident: IncidentResponse | null | undefined
+): DecisionChainDisplay | null {
+  if (!incident?.decision?.proposal_data) {
+    return null
+  }
+
+  const data = incident.decision.proposal_data
+  const target = incident.affected_services?.[0] || 'unknown-service'
+  const source = data.source || 'unknown'
+  const now = new Date().toISOString()
+
+  // 建構推理步驟 (從 API 資料)
+  const steps: ReasoningStep[] = [
+    {
+      step: 'SIGNAL_RECEIVED',
+      reasoning: `收到 ${incident.signal_count || 1} 筆告警，影響服務: ${target}`,
+      timestamp: incident.created_at,
+    },
+    {
+      step: 'SEVERITY_EVALUATION',
+      reasoning: `告警等級: ${incident.severity} | 狀態: ${incident.status}`,
+    },
+    {
+      step: 'AI_ENGINE',
+      reasoning: `決策引擎: ${source === 'expert_system' ? 'Expert System (規則引擎)' : 'OpenClaw LLM (智能分析)'}`,
+    },
+  ]
+
+  // 加入 AI 推理 (如果有)
+  if (data.reasoning) {
+    steps.push({
+      step: 'ROOT_CAUSE_ANALYSIS',
+      reasoning: data.reasoning,
+    })
+  }
+
+  // 加入描述 (如果有)
+  if (data.description) {
+    steps.push({
+      step: 'ANALYSIS_RESULT',
+      reasoning: data.description,
+    })
+  }
+
+  // 加入建議動作
+  if (data.action || data.kubectl_command) {
+    steps.push({
+      step: 'ACTION_RECOMMENDATION',
+      reasoning: `建議動作: ${data.action || data.kubectl_command}\n風險等級: ${data.risk_level || 'medium'}`,
+    })
+  }
+
+  return {
+    analysis_type: 'action_suggestion',
+    target_service: target,
+    reasoning_steps: steps,
+    conclusion: data.action || '等待 AI 分析完成...',
+    confidence: data.confidence ?? 0.75,
+  }
+}
+
 // =============================================================================
 // Utility: Map IncidentResponse to DualStateIncidentCard props
 // =============================================================================
@@ -236,13 +319,17 @@ export default function Home({ params }: { params: { locale: string } }) {
             )}
           </DataPincerPanel>
 
-          {/* OpenClaw Thinking Terminal (Phase 7: 決策鏈視覺化) */}
+          {/* OpenClaw Thinking Terminal (Phase 7: 決策鏈視覺化 - 真實 AI 資料) */}
           <DataPincerPanel
             title="OpenClaw Terminal"
-            status="thinking"
+            status={(incidents?.length || 0) > 0 ? "thinking" : "healthy"}
           >
             <ThinkingTerminal
-              decisionChain={(incidents?.length || 0) > 0 ? DEMO_DECISION_CHAIN : null}
+              decisionChain={
+                (incidents?.length || 0) > 0
+                  ? convertToDecisionChain(incidents?.[0])
+                  : null
+              }
               incidentId={(incidents?.length || 0) > 0 ? incidents?.[0]?.incident_id : undefined}
               autoPlay={(incidents?.length || 0) > 0}
               maxHeight="300px"