myagent-ai 1.16.13 → 1.16.14

package/agents/main_agent.py

@@ -693,6 +693,41 @@ class MainAgent(BaseAgent):
                         )
                     break
 
+                # [v1.16.13] 特殊处理模型不支持图片输入 — 去掉图片用纯文本重试
+                _vision_keywords = ["doesn't support image", "does not support image", "model_incompatible", "image input", "not support vision", "unsupported multimodal", "image capability"]
+                if any(kw in _llm_error.lower() for kw in _vision_keywords) and context.metadata.get("user_images"):
+                    logger.warning(f"[{task_id}] 模型不支持图片输入，去掉图片用纯文本重试")
+                    context.metadata["user_images"] = []
+                    # 用纯文本消息替换最后一条多模态消息
+                    _text_only_msg = context.user_message or "请处理上述上下文。"
+                    if len(messages) > 0 and isinstance(messages[-1].content, list):
+                        messages[-1] = Message(role="user", content=_text_only_msg)
+                    # 重试 LLM 调用
+                    if stream_response and self.llm:
+                        response = await self._call_llm_stream(
+                            messages, text_delta_callback=text_delta_callback,
+                            stream_response=stream_response,
+                        )
+                    else:
+                        response = await self._call_llm(messages)
+                    if response.success:
+                        # 纯文本重试成功，给回复加上提示前缀
+                        _vision_prefix = "⚠️ 当前模型不支持图片识别，已自动使用纯文本模式处理（图片未发送给模型）。\n\n"
+                        llm_raw = _vision_prefix + response.content
+                        context.working_memory["final_response"] = llm_raw
+                        await self._emit_v2_event("v2_reasoning", {"content": llm_raw}, stream_callback)
+                        if self.memory:
+                            self.memory.add_session(
+                                session_id=context.session_id,
+                                role="assistant",
+                                content=llm_raw,
+                            )
+                        break
+                    else:
+                        # 纯文本也失败了，走下面的通用错误处理
+                        _llm_error = response.error or ""
+                        logger.error(f"[{task_id}] 纯文本重试也失败: {_llm_error}")
+
                 # 其他 LLM 错误
                 error_msg = f"LLM 调用失败: {response.error}"
                 context.working_memory["final_response"] = error_msg

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "myagent-ai",
-  "version": "1.16.13",
+  "version": "1.16.14",
   "description": "本地桌面端执行型AI助手 - Open Interpreter 风格 | Local Desktop Execution-Oriented AI Assistant",
   "main": "main.py",
   "bin": {
@@ -43,4 +43,4 @@
     "python": ">=3.10",
     "node": ">=18"
   }
-}
+}