feat(engine): 添加MCP连接管理和工具集成功能

- 集成MCP连接管理器，支持MCP服务器连接
- 添加多种内置工具：ClarifyTool、CronTool、DelegateTool、ExecuteCodeTool、
  PatchFileTool、ProcessTool、SendMessageTool、SpawnTool、TerminalTool、
  TodoTool、WebFetchTool、WebSearchTool、WriteFileTool等
- 实现工具注册和装配功能
- 添加技能选择上下文参数
- 支持思考模式控制参数thinking_enabled

feat(coordinator): 重构任务执行计划器参数命名

- 将learning_candidate_enabled重命名为allow_candidate_generation
- 更新TeamGraphScheduler中的参数传递
- 修改LocalAgentRunner中的相关参数处理
- 更新README文档中的相应描述

refactor(context): 标准化工具调用参数格式

- 添加_json导入用于参数序列化
- 实现_provider_tool_calls方法标准化OpenAI兼容的工具调用载荷
- 修复工具调用中参数非字符串类型的序列化问题

refactor(session): 优化消息历史记录过滤逻辑

- 修改get_messages_as_conversation为基于运行状态过滤消息
- 排除未完成、失败或错误结束的运行记录
- 改进对话历史的可见性控制机制

fix(store): 修复FTS索引重建逻辑

- 添加异常处理防止FTS索引创建失败
- 实现_rebuild_fts_index方法重新构建全文搜索索引
- 优化索引触发器和表的维护流程

app-instance/backend/beaver/engine/providers/litellm.py

@@ -123,6 +123,25 @@ class LiteLLMProvider(LLMProvider):
             clean = {key: value for key, value in message.items() if key in _ALLOWED_MSG_KEYS}
             if clean.get("role") == "assistant" and "content" not in clean:
                 clean["content"] = None
+            if isinstance(clean.get("tool_calls"), list):
+                clean["tool_calls"] = LiteLLMProvider._sanitize_tool_calls(clean["tool_calls"])
+            sanitized.append(clean)
+        return sanitized
+
+    @staticmethod
+    def _sanitize_tool_calls(tool_calls: list[Any]) -> list[dict[str, Any]]:
+        sanitized: list[dict[str, Any]] = []
+        for tool_call in tool_calls:
+            if not isinstance(tool_call, dict):
+                continue
+            clean = dict(tool_call)
+            function = clean.get("function")
+            if isinstance(function, dict):
+                clean_function = dict(function)
+                arguments = clean_function.get("arguments")
+                if not isinstance(arguments, str):
+                    clean_function["arguments"] = json.dumps(arguments or {}, ensure_ascii=False, default=str)
+                clean["function"] = clean_function
             sanitized.append(clean)
         return sanitized
 
@@ -155,6 +174,18 @@ class LiteLLMProvider(LLMProvider):
         if provider_payload:
             kwargs["provider"] = provider_payload
 
+    def _apply_thinking_mode(self, original_model: str, resolved_model: str, kwargs: dict[str, Any], enabled: bool | None) -> None:
+        if enabled is None:
+            return
+        model_key = f"{original_model} {resolved_model}".lower()
+        if "qwen" not in model_key:
+            return
+        extra_body = dict(kwargs.get("extra_body") or {})
+        chat_template_kwargs = dict(extra_body.get("chat_template_kwargs") or {})
+        chat_template_kwargs["enable_thinking"] = bool(enabled)
+        extra_body["chat_template_kwargs"] = chat_template_kwargs
+        kwargs["extra_body"] = extra_body
+
     async def chat(
         self,
         messages: list[dict[str, Any]],
@@ -162,6 +193,7 @@ class LiteLLMProvider(LLMProvider):
         model: str | None = None,
         max_tokens: int = 4096,
         temperature: float = 0.7,
+        thinking_enabled: bool | None = None,
     ) -> LLMResponse:
         if acompletion is None:
             return LLMResponse(content="Error: litellm is not installed", finish_reason="error", provider_name=self.provider_name)
@@ -174,6 +206,7 @@ class LiteLLMProvider(LLMProvider):
             "messages": sanitized_messages,
             "max_tokens": max(1, max_tokens),
             "temperature": temperature,
+            "timeout": self.request_timeout_seconds or 45.0,
         }
         if self.api_key:
             kwargs["api_key"] = self.api_key
@@ -186,6 +219,7 @@ class LiteLLMProvider(LLMProvider):
             kwargs["tool_choice"] = "auto"
         self._apply_model_overrides(original_model, kwargs)
         self._apply_openrouter_routing(kwargs)
+        self._apply_thinking_mode(original_model, resolved_model, kwargs, thinking_enabled)
         env_overrides = self._build_env_overrides(self.api_key, self.api_base, original_model)
 
         try: