Trying to fix non-streaming

trustgraph-base/trustgraph/api/socket_client.py

@@ -118,32 +118,20 @@ class SocketClient:
         async with websockets.connect(ws_url, ping_interval=20, ping_timeout=self.timeout) as websocket:
             await websocket.send(json.dumps(message))
 
-            # Some services (like agent) send multiple messages even in non-streaming mode
+            # Wait for single response
-            # Collect all messages until complete=True
+            raw_message = await websocket.recv()
-            messages = []
+            response = json.loads(raw_message)
 
-            while True:
+            if response.get("id") != request_id:
-                raw_message = await websocket.recv()
+                raise ProtocolException(f"Response ID mismatch")
-                response = json.loads(raw_message)
 
-                if response.get("id") != request_id:
+            if "error" in response:
-                    raise ProtocolException(f"Response ID mismatch")
+                raise_from_error_dict(response["error"])
 
-                if "error" in response:
+            if "response" not in response:
-                    raise_from_error_dict(response["error"])
+                raise ProtocolException(f"Missing response in message")
 
-                if "response" not in response:
+            return response["response"]
-                    raise ProtocolException(f"Missing response in message")
-
-                messages.append(response["response"])
-
-                # Check if this is the final message
-                if response.get("complete", False):
-                    break
-
-            # For single-message responses, return the dict directly
-            # For multi-message responses (like agent), return a list
-            return messages[0] if len(messages) == 1 else messages
 
     async def _send_request_async_streaming(
         self,
@@ -226,6 +214,23 @@ class SocketClient:
                 content=resp.get("content", ""),
                 end_of_message=resp.get("end_of_message", False)
             )
+        # Non-streaming agent format: chunk_type is empty but has thought/observation/answer fields
+        elif resp.get("thought"):
+            return AgentThought(
+                content=resp.get("thought", ""),
+                end_of_message=resp.get("end_of_message", False)
+            )
+        elif resp.get("observation"):
+            return AgentObservation(
+                content=resp.get("observation", ""),
+                end_of_message=resp.get("end_of_message", False)
+            )
+        elif resp.get("answer"):
+            return AgentAnswer(
+                content=resp.get("answer", ""),
+                end_of_message=resp.get("end_of_message", False),
+                end_of_dialog=resp.get("end_of_dialog", False)
+            )
         else:
             # RAG-style chunk (or generic chunk)
             # Text-completion uses "response" field, RAG uses "chunk" field, Prompt uses "text" field
@@ -273,7 +278,9 @@ class SocketFlowInstance:
             request["history"] = history
         request.update(kwargs)
 
-        return self.client._send_request_sync("agent", self.flow_id, request, streaming)
+        # Agents always use multipart messaging (multiple complete messages)
+        # regardless of streaming flag, so always use the streaming code path
+        return self.client._send_request_sync("agent", self.flow_id, request, streaming=True)
 
     def text_completion(self, system: str, prompt: str, streaming: bool = False, **kwargs) -> Union[str, Iterator[str]]:
         """Text completion with optional streaming"""

trustgraph-cli/trustgraph/cli/invoke_agent.py

@@ -178,28 +178,22 @@ def question(
                 print()
 
         else:
-            # Non-streaming response
+            # Non-streaming response - but agents use multipart messaging
-            # Response may be a list of messages (thought/observation/answer) or a single dict
+            # so we iterate through the chunks (which are complete messages, not text chunks)
-            messages = response if isinstance(response, list) else [response]
+            for chunk in response:
-
-            for msg in messages:
                 # Display thoughts if verbose
-                if "thought" in msg and msg["thought"] and verbose:
+                if chunk.chunk_type == "thought" and verbose:
-                    output(wrap(msg["thought"]), "\U0001f914 ")
+                    output(wrap(chunk.content), "\U0001f914 ")
                     print()
 
                 # Display observations if verbose
-                if "observation" in msg and msg["observation"] and verbose:
+                elif chunk.chunk_type == "observation" and verbose:
-                    output(wrap(msg["observation"]), "\U0001f4a1 ")
+                    output(wrap(chunk.content), "\U0001f4a1 ")
                     print()
 
                 # Display answer
-                if "answer" in msg and msg["answer"]:
+                elif chunk.chunk_type == "final-answer" or chunk.chunk_type == "answer":
-                    print(msg["answer"])
+                    print(chunk.content)
-
-                # Handle errors
-                if "error" in msg:
-                    raise RuntimeError(msg["error"])
 
     finally:
         # Clean up socket connection