put base64 in the currect position for sb_vision

Author: CZH-THU

app/agent/sandbox_agent.py

@@ -11,7 +11,6 @@
 from app.prompt.manus import NEXT_STEP_PROMPT, SYSTEM_PROMPT
 from app.tool import Terminate, ToolCollection
 from app.tool.ask_human import AskHuman
-from app.tool.browser_use_tool import BrowserUseTool
 from app.tool.mcp import MCPClients, MCPClientTool
 from app.tool.sandbox.sb_browser_tool import SandboxBrowserTool
 from app.tool.sandbox.sb_files_tool import SandboxFilesTool
@@ -205,7 +204,7 @@ async def think(self) -> bool:
         original_prompt = self.next_step_prompt
         recent_messages = self.memory.messages[-3:] if self.memory.messages else []
         browser_in_use = any(
-            tc.function.name == BrowserUseTool().name
+            tc.function.name == SandboxBrowserTool().name
             for msg in recent_messages
             if msg.tool_calls
             for tc in msg.tool_calls

app/daytona/sandbox.py

@@ -1,3 +1,5 @@
+import time
+
 from daytona import (
     CreateSandboxFromImageParams,
     Daytona,
@@ -90,6 +92,7 @@ def start_supervisord_session(sandbox: Sandbox):
                 var_async=True,
             ),
         )
+        time.sleep(25)  # Wait a bit to ensure supervisord starts properly
         logger.info(f"Supervisord started in session {session_id}")
     except Exception as e:
         logger.error(f"Error starting supervisord session: {str(e)}")

app/tool/sandbox/sb_browser_tool.py

@@ -229,7 +229,7 @@ async def _execute_browser_action(
                                 f"Screenshot validation failed: {validation_message}"
                             )
                             result["image_validation_error"] = validation_message
-                        del result["screenshot_base64"]
+                            del result["screenshot_base64"]
 
                     # added_message = await self.thread_manager.add_message(
                     #     thread_id=self.thread_id,

app/tool/sandbox/sb_vision_tool.py

@@ -170,8 +170,9 @@ async def execute(
             )
             self.vision_message = message
             # return self.success_response(f"成功加载并压缩图片 '{cleaned_path}' (由 {file_info.size / 1024:.1f}KB 压缩到 {len(compressed_bytes) / 1024:.1f}KB)。")
-            return self.success_response(
-                f"成功加载并压缩图片 '{cleaned_path}'，压缩后的内容为：{base64_image}"
+            return ToolResult(
+                output=f"成功加载并压缩图片 '{cleaned_path}'",
+                base64_image=base64_image,
             )
         except Exception as e:
             return self.fail_response(f"see_image 执行异常: {str(e)}")