From ee2026a97388b9c676ea2773c89fae1f0494bda2 Mon Sep 17 00:00:00 2001
From: Your Name <your.email@example.com>
Date: Thu, 19 Feb 2026 02:59:50 +0800
Subject: [PATCH] fix(agent): tolerate non-json analyzer replies and cap graph
 step latency

---
 lib/ai/auto-agent-graph.ts | 40 +++++++++++++++++++++++++++++++++++++-
 lib/ai/smart-analyzer.ts   | 34 ++++++++++++++++++++++++++++++--
 2 files changed, 71 insertions(+), 3 deletions(-)

diff --git a/lib/ai/auto-agent-graph.ts b/lib/ai/auto-agent-graph.ts
index 7b5ae4986..470c412c9 100644
--- a/lib/ai/auto-agent-graph.ts
+++ b/lib/ai/auto-agent-graph.ts
@@ -85,6 +85,24 @@ function compactText(text: string, max = 48) {
   return `${normalized.slice(0, max)}...`;
 }
 
+async function withTimeoutFallback<T>(
+  promise: Promise<T>,
+  timeoutMs: number,
+  fallbackValue: T,
+): Promise<T> {
+  let timeoutId: NodeJS.Timeout | undefined;
+  try {
+    return await Promise.race([
+      promise,
+      new Promise<T>((resolve) => {
+        timeoutId = setTimeout(() => resolve(fallbackValue), timeoutMs);
+      }),
+    ]);
+  } finally {
+    if (timeoutId) clearTimeout(timeoutId);
+  }
+}
+
 const AGENT_VISION_MAX_IMAGES = 4;
 const AGENT_VISION_MAX_SIDE = 500;
 const AGENT_VISION_FETCH_TIMEOUT_MS = 10_000;
@@ -484,6 +502,7 @@ function buildStrategicFriendlyMessage(
   state: AutoAgentStateType,
 ): Promise<string | null> {
   const fallbackQuestion = analysis.clarifyingQuestions?.[0];
+  const existingReply = analysis.conversationalReply?.trim();
 
   if (analysis.needsClarification) {
     return Promise.resolve(
@@ -491,7 +510,21 @@ function buildStrategicFriendlyMessage(
     );
   }
 
-  return generateAgentConversation(analysis, state);
+  if (existingReply) {
+    const isStructuredReply =
+      /###\s*需求理解/.test(existingReply) ||
+      /###\s*图片分析/.test(existingReply) ||
+      /###\s*执行计划/.test(existingReply);
+    if (isStructuredReply || existingReply.length >= 24) {
+      return Promise.resolve(existingReply);
+    }
+  }
+
+  return withTimeoutFallback(
+    generateAgentConversation(analysis, state),
+    12000,
+    existingReply || null,
+  );
 }
 
 type EnabledModelWithProvider = Awaited<
@@ -694,6 +727,7 @@ async function generateAgentConversation(
           ],
           temperature: 0.2,
           maxTokens: 360,
+          abortSignal: AbortSignal.timeout(12000),
         } as any);
 
         visualAnalysis = visionText.trim();
@@ -746,6 +780,7 @@ async function generateAgentConversation(
       prompt,
       temperature: 0.35,
       maxTokens: 520,
+      abortSignal: AbortSignal.timeout(12000),
     } as any);
 
     const cleaned = text.trim();
@@ -819,6 +854,7 @@ async function recommendImageParamsWithAgent(
       prompt: requestText,
       temperature: 0.15,
       maxTokens: 220,
+      abortSignal: AbortSignal.timeout(10000),
     } as any);
 
     const parsed = parseJsonBlock<AgentRecommendedImageParams>(text);
@@ -903,6 +939,7 @@ async function recommendVideoPlanWithAgent(
       prompt,
       temperature: 0.2,
       maxTokens: 720,
+      abortSignal: AbortSignal.timeout(12000),
     } as any);
 
     const parsed = parseJsonBlock<Record<string, unknown>>(text);
@@ -974,6 +1011,7 @@ async function refineRouteWithAgent(
 ${candidatesText}`,
       temperature: 0.1,
       maxTokens: 120,
+      abortSignal: AbortSignal.timeout(8000),
     } as any);
 
     const parsed = parseJsonBlock<{ modelId?: string; reason?: string }>(text);
diff --git a/lib/ai/smart-analyzer.ts b/lib/ai/smart-analyzer.ts
index d892cd77c..19133a96e 100644
--- a/lib/ai/smart-analyzer.ts
+++ b/lib/ai/smart-analyzer.ts
@@ -373,7 +373,21 @@ function parseAnalysisResultFromModelText(
   const jsonMatch = responseText.match(/\{[\s\S]*\}/);
   if (!jsonMatch) {
     console.warn("[Smart Analyzer] Model response missing JSON:", responseText.slice(0, 240));
-    return null;
+    const fallback = fallbackKeywordAnalysis(userText, hasImage);
+    const conversationalReply = responseText.trim();
+    if (!conversationalReply) {
+      return null;
+    }
+    return normalizeIntentForImageContext(
+      {
+        ...fallback,
+        confidence: Math.max(fallback.confidence, 0.82),
+        needsClarification: false,
+        conversationalReply,
+      },
+      userText,
+      hasImage,
+    );
   }
 
   try {
@@ -386,7 +400,21 @@ function parseAnalysisResultFromModelText(
       "raw=",
       jsonMatch[0].slice(0, 240),
     );
-    return null;
+    const fallback = fallbackKeywordAnalysis(userText, hasImage);
+    const conversationalReply = responseText.trim();
+    if (!conversationalReply) {
+      return null;
+    }
+    return normalizeIntentForImageContext(
+      {
+        ...fallback,
+        confidence: Math.max(fallback.confidence, 0.8),
+        needsClarification: false,
+        conversationalReply,
+      },
+      userText,
+      hasImage,
+    );
   }
 }
 
@@ -902,6 +930,7 @@ ${historyText}
         ],
         temperature: 0.1,
         maxTokens: 360,
+        abortSignal: AbortSignal.timeout(12000),
       }
     : {
         model,
@@ -909,6 +938,7 @@ ${historyText}
         prompt: userMessage,
         temperature: 0.1,
         maxTokens: 300,
+        abortSignal: AbortSignal.timeout(10000),
       };
 
   const { text: responseText } = await generateText(requestPayload as any);