microsoft/autogen · error · InvalidOperationException

Invalid streaming message type {reply.GetType().Name}

Error message

Invalid streaming message type {reply.GetType().Name}

What it means

OpenAIChatRequestMessageConnector's streaming path only processes MessageEnvelope<StreamingChatCompletionsUpdate> chunks (text deltas and tool-call deltas). Any other streaming message type hits this branch; in strictMode it throws InvalidOperationException, otherwise it passes the message through untouched.

Source

Thrown at dotnet/src/AutoGen.OpenAI.V1/Middleware/OpenAIChatRequestMessageConnector.cs:80

                if (update.Content.FunctionName is string functionName)
                {
                    currentToolName = functionName;
                }
                else if (update.Content.ToolCallUpdate is StreamingFunctionToolCallUpdate toolCallUpdate && toolCallUpdate.Name is string toolCallName)
                {
                    currentToolName = toolCallName;
                }
                var postProcessMessage = PostProcessStreamingMessage(update, currentToolName);
                if (postProcessMessage != null)
                {
                    yield return postProcessMessage;
                }
            }
            else
            {
                if (this.strictMode)
                {
                    throw new InvalidOperationException($"Invalid streaming message type {reply.GetType().Name}");
                }
                else
                {
                    yield return reply;
                }
            }
        }
    }

    public IMessage PostProcessMessage(IMessage message)
    {
        return message switch
        {
            IMessage<ChatResponseMessage> m => PostProcessChatResponseMessage(m.Content, m.From),
            IMessage<ChatCompletions> m => PostProcessChatCompletions(m),
            _ when strictMode is false => message,
            _ => throw new InvalidOperationException($"Invalid return message type {message.GetType().Name}"),
        };

View on GitHub (pinned to 027ecf0a37)

Solutions

1. Construct the connector with strictMode: false (default) so foreign streaming messages pass through
2. Ensure only MessageEnvelope<StreamingChatCompletionsUpdate> items flow through the OpenAI connector's streaming path, or place custom middleware after it
3. Have custom middleware wrap its output as updates or defer emission to post-processing

Example fix

// before
var connector = new OpenAIChatRequestMessageConnector(strictMode: true);
agent = agent.RegisterStreamingMiddleware(connector)
              .RegisterStreamingMiddleware(myCustomYieldingMiddleware); // throws in strict mode

// after
var connector = new OpenAIChatRequestMessageConnector(strictMode: false);
agent = agent.RegisterStreamingMiddleware(connector)
              .RegisterStreamingMiddleware(myCustomYieldingMiddleware);

bool isUpdate = reply is MessageEnvelope<StreamingChatCompletionsUpdate>;
if (!isUpdate && connectorStrictMode) { /* route around connector */ }

static bool IsStreamingUpdate(IMessage m) =>
    m is MessageEnvelope<StreamingChatCompletionsUpdate>;

catch (InvalidOperationException ex) when (ex.Message.Contains("Invalid streaming message type"))
{
    logger.LogError("Custom streaming middleware clashed with strict connector; disabling strict mode");
    connector = new OpenAIChatRequestMessageConnector(strictMode: false);
}

When it happens

Trigger: Stacking middlewares that inject custom IMessage types into the streaming pipeline ahead of this connector while it was constructed with strictMode: true, then invoking GenerateStreamingReplyAsync.

Common situations: Composing custom streaming middleware (logging, routing, caching) that yields non-update messages; mixing connectors from different provider packages on one agent with strict mode enabled.

Related errors

• Invalid streaming message type {reply.GetType().Name}
• Invalid return message type {message.GetType().Name}
• Please set OPENAI_API_KEY environment variable.
• The first message is ToolCallMessage, but the update message
• The message is not a valid message

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/58698494d206f92e. Report an issue: GitHub.