Fix: agent toolcall null response & schema validation & DeepSeek think history (#14425)

### What problem does this PR solve?
agent toolcall null response & schema validation & DeepSeek think
history

### Type of change

- [x] Bug Fix (non-breaking change which fixes an issue)

agent/component/agent_with_tools.py

@@ -145,7 +145,8 @@ class Agent(LLM, ToolBase):
         self._param.function_name = self._id.split("-->")[-1]
         m = super().get_meta()
         if hasattr(self._param, "user_prompt") and self._param.user_prompt:
-            m["function"]["parameters"]["properties"]["user_prompt"] = self._param.user_prompt
+            # Keep the JSON schema valid; user_prompt is a string field, not a schema node.
+            m["function"]["parameters"]["properties"]["user_prompt"]["default"] = self._param.user_prompt
         return m
 
     def get_input_form(self) -> dict[str, dict]:

agent/tools/base.py

@@ -67,6 +67,19 @@ class LLMToolPluginCallSession(ToolCallSession):
         else:
             resp = await thread_pool_exec(tool_obj.invoke, **arguments)
 
+        if resp is None and hasattr(tool_obj, "output") and callable(tool_obj.output):
+            try:
+                fallback_output = tool_obj.output()
+                if isinstance(fallback_output, dict) and fallback_output.get("content") not in (None, ""):
+                    resp = fallback_output["content"]
+                elif fallback_output not in (None, ""):
+                    resp = fallback_output
+                else:
+                    resp = fallback_output
+                logging.warning(f"[ToolCall] resp is None, fallback to output name={name} output_keys={list(fallback_output.keys()) if isinstance(fallback_output, dict) else type(fallback_output).__name__}")
+            except Exception as e:
+                logging.warning(f"[ToolCall] resp is None and output fallback failed name={name} err={e}")
+
         elapsed = timer() - st
         logging.info(f"[ToolCall] done name={name} elapsed={elapsed:.2f}s result={str(resp)[:200]}")
         self.callback(name, arguments, resp, elapsed_time=elapsed)