feat: 提炼模板弹窗加宽,后端配置上下文压缩测试值

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

frontend-web/src/components/wiki/KnowledgeManage.tsx

@@ -304,7 +304,7 @@ export function ExtractTemplateManageDialog({
 
   return (
     <Dialog open={open} onOpenChange={onOpenChange}>
-      <DialogContent className="max-w-4xl">
+      <DialogContent className="w-[95vw] max-w-[95vw] sm:max-w-[1800px]">
         <DialogHeader>
           <DialogTitle>抽取提示词模板</DialogTitle>
           <DialogDescription>

offline-backend-20260512/backend/config.yaml

@@ -190,6 +190,28 @@ title:
     User: {user_msg}
     Assistant: {assistant_msg}
 
+# 上下文压缩(对话摘要):长对话接近上下文上限时,把较早的消息总结成一段摘要,
+# 只保留最近若干条原始消息,避免撑爆模型上下文窗口。
+# - trigger:触发条件,可配多个,任意一个先满足就压缩(OR 关系)。真正决定何时
+#     压缩的是 trigger(不是 keep)。type=tokens 按 token 总数;type=messages 按条数。
+# - keep:压缩后保留多少最近的上下文(tokens / messages)。
+# - trim_tokens_to_summarize:喂给摘要模型的消息最多裁到多少 token。
+# 当前为「测试值(易触发)」:发到服务器后稍微聊几句就会触发,可在 Gateway 日志里
+# grep「上下文压缩」/「context compaction」确认生效;确认后再调回生产值
+# (trigger tokens 60000 / messages 80,keep tokens 16000)。
+summarization:
+  enabled: true
+  # model_name: deepseek-chat   # 留空则用主对话模型
+  trigger:
+    - type: tokens
+      value: 8000
+    - type: messages
+      value: 12
+  keep:
+    type: messages
+    value: 6
+  trim_tokens_to_summarize: 4000
+
 tool_groups:
   - name: web
   - name: file:read

offline-backend-20260512/config.yaml

@@ -307,9 +307,9 @@ summarization:
   # model_name: deepseek-chat
   trigger:
     - type: tokens
-      value: 60000
+      value: 8000
     - type: messages
-      value: 80
+      value: 12
   # Keep a small *token* budget of recent context (same unit as the token
   # trigger, well below it). This is what the model receives after compaction:
   # [summary, ~16000 tokens of recent turns]. Small keep = small model context
@@ -318,6 +318,9 @@ summarization:
   # which spans many turns. The full conversation still stays in state for the
   # page; only what is sent to the model is compressed.
   keep:
-    type: tokens
-    value: 16000
+    type: messages   # 测试值:压缩后只保留最近 6 条原始消息
+    value: 6
+    # 生产值改用 token 预算:
+    # type: tokens
+    # value: 16000
   trim_tokens_to_summarize: 4000