.NET: [Bug]: invalid_payload after the AI responeded with reasoning role.

[karljucutan]

Description

After I added a ReasoningOptions, I got invalid_payload error after the AI responded a message with role: "reasoning". If the AI did not respond yet with role: "reasoning" there are no invalid_payload issue.

Request Payload:

{
  "threadId": "thread-1783146160899-7dx2yk",
  "runId": "run-1783146211845-bxzbae",
  "state": {},
  "messages": [
    {
      "id": "msg-1783146164467-7cvmjc",
      "role": "user",
      "parts": [
        {
          "type": "text",
          "content": "hi"
        }
      ],
      "createdAt": "2026-07-04T06:22:44.467Z",
      "content": "hi"
    },
    {
      "id": "msg_08401b0393b4b483006a48a6bac59881958aa78143e8490bfe",
      "role": "assistant",
      "parts": [
        {
          "type": "text",
          "content": "Hi — how can I help you today? I can search business rules, explain rules, map requirements to rules, or create acceptance criteria and test cases. Which would you like?"
        }
      ],
      "createdAt": "2026-07-04T06:22:50.154Z",
      "content": "Hi — how can I help you today? I can search business rules, explain rules, map requirements to rules, or create acceptance criteria and test cases. Which would you like?"
    },
    {
      "id": "msg-1783146176003-b16cnh",
      "role": "user",
      "parts": [
        {
          "type": "text",
          "content": "reason to me"
        }
      ],
      "createdAt": "2026-07-04T06:22:56.003Z",
      "content": "reason to me"
    },
    {
      "role": "reasoning",
      "id": "msg_0b3cc0809fa6331c006a48a6c73d80819399a7d1514a02b948-reasoning-rx7ptq",
      "content": "**Clarifying business rules**\n\nI’m wondering if the user wants me to explain my reasoning about a specific business rule or decision. If that's the case, I need to know which rule or scenario they mean. It’s important to remember that I must use SearchRules to find applicable business rules if needed, and I should keep my questions concise, especially when I’m unsure. Keeping things short and clear is key!"
    },
    {
      "id": "msg_0b3cc0809fa6331c006a48a6c73d80819399a7d1514a02b948",
      "role": "assistant",
      "parts": [
        {
          "type": "thinking",
          "content": "**Clarifying business rules**\n\nI’m wondering if the user wants me to explain my reasoning about a specific business rule or decision. If that's the case, I need to know which rule or scenario they mean. It’s important to remember that I must use SearchRules to find applicable business rules if needed, and I should keep my questions concise, especially when I’m unsure. Keeping things short and clear is key!",
          "stepId": null
        },
        {
          "type": "text",
          "content": "Do you mean you want me to explain my reasoning about a specific business rule or scenario — if so, which rule ID or scenario should I reason about?"
        }
      ],
      "createdAt": "2026-07-04T06:23:01.673Z",
      "content": "Do you mean you want me to explain my reasoning about a specific business rule or scenario — if so, which rule ID or scenario should I reason about?"
    },
    {
      "id": "msg-1783146211845-17rr6i",
      "role": "user",
      "parts": [
        {
          "type": "text",
          "content": "hi"
        }
      ],
      "createdAt": "2026-07-04T06:23:31.845Z",
      "content": "hi"
    }
  ],
  "tools": [],
  "context": [],
  "forwardedProps": {},
  "data": {}
}

EventStream:

{"threadId":"thread-1783146160899-7dx2yk","runId":"run-1783146211845-bxzbae","type":"RUN_STARTED"}
{"message":"HTTP 400 (ServiceError: invalid_payload)\r\n\r\nThe provided data does not match the expected schema [Request ID: caefc2e768a37cbab6286b5fda3d550c]\r\n\r\n{\r\n  \u0022code\u0022: \u0022invalid_payload\u0022,\r\n  \u0022message\u0022: \u0022The provided data does not match the expected schema [Request ID: caefc2e768a37cbab6286b5fda3d550c]\u0022,\r\n  \u0022param\u0022: \u0022/\u0022,\r\n  \u0022type\u0022: \u0022invalid_request_error\u0022,\r\n  \u0022details\u0022: [],\r\n  \u0022additionalInfo\u0022: {\r\n    \u0022request_id\u0022: \u0022caefc2e768a37cbab6286b5fda3d550c\u0022\r\n  }\r\n}","code":"StreamingError","type":"RUN_ERROR"}

Code Sample

using Azure.AI.Projects;
using Azure.Identity;
using Microsoft.Agents.AI.Hosting;
using Microsoft.Extensions.AI;

namespace API.Features.Rules.MafAgents;

public static class SearchRulesAgentBuilderExtensions
{
    public static WebApplicationBuilder AddRuleAssistantAIAgent(this WebApplicationBuilder builder)
    {
        string endpoint = builder.Configuration["AZURE_OPENAI_ENDPOINT"]
            ?? throw new InvalidOperationException("AZURE_OPENAI_ENDPOINT is not set.");
        string deploymentName = builder.Configuration["AZURE_OPENAI_DEPLOYMENT_NAME"]
            ?? throw new InvalidOperationException("AZURE_OPENAI_DEPLOYMENT_NAME is not set.");

        builder.Services.AddSingleton<SearchRulesAgentTool>();

        builder.AddAIAgent(
            name: "RulesAssistantAIAgent",
            (serviceProvider, key) =>
            {
                var tool = serviceProvider.GetRequiredService<SearchRulesAgentTool>();

                return new AIProjectClient(new Uri(endpoint), new DefaultAzureCredential())
                    .AsAIAgent(
                        model: deploymentName,
                        name: key,
                        instructions: @"You are an expert Business Rules Assistant.",
                        tools: [AIFunctionFactory.Create(tool.ExecuteSearchRulesTool)],
                        clientFactory: innerClient => innerClient
                            .AsBuilder()
                            .ConfigureOptions(options =>
                            {
                                // Forces gpt-5-mini to output its internal thought chain
                                options.Reasoning = new ReasoningOptions
                                {
                                    Effort = ReasoningEffort.Medium,
                                    Output = ReasoningOutput.Full
                                };
                            })
                            .Build(),
                        services: serviceProvider // Gives the tool access to scoped DI dependencies
                    );

            });

        return builder;
    }
}

Error Messages / Stack Traces

Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIServerSentEventsResult: Error: An error occurred while streaming AG-UI events

System.ClientModel.ClientResultException: HTTP 400 (ServiceError: invalid_payload)

The provided data does not match the expected schema [Request ID: caefc2e768a37cbab6286b5fda3d550c]

{
  "code": "invalid_payload",
  "message": "The provided data does not match the expected schema [Request ID: caefc2e768a37cbab6286b5fda3d550c]",
  "param": "/",
  "type": "invalid_request_error",
  "details": [],
  "additionalInfo": {
    "request_id": "caefc2e768a37cbab6286b5fda3d550c"
  }
}
   at OpenAI.ClientPipelineExtensions.ProcessMessageAsync(ClientPipeline pipeline, PipelineMessage message, RequestOptions options)
   at OpenAI.Responses.ResponsesClient.CreateResponseAsync(BinaryContent content, RequestOptions options)
   at OpenAI.Responses.ResponsesClient.<>c__DisplayClass18_0.<<CreateResponseStreamingAsync>b__0>d.MoveNext()
--- End of stack trace from previous location ---
   at OpenAI.AsyncSseUpdateCollection`1.GetRawPagesAsync()+MoveNext()
   at OpenAI.AsyncSseUpdateCollection`1.GetRawPagesAsync()+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at System.ClientModel.AsyncCollectionResult`1.GetAsyncEnumerator(CancellationToken cancellationToken)+MoveNext()
   at System.ClientModel.AsyncCollectionResult`1.GetAsyncEnumerator(CancellationToken cancellationToken)+MoveNext()
   at System.ClientModel.AsyncCollectionResult`1.GetAsyncEnumerator(CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Extensions.AI.OpenAIResponsesChatClient.FromOpenAIStreamingResponseUpdatesAsync(IAsyncEnumerable`1 streamingResponseUpdates, CreateResponseOptions options, String conversationId, String resumeResponseId, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.OpenAIResponsesChatClient.FromOpenAIStreamingResponseUpdatesAsync(IAsyncEnumerable`1 streamingResponseUpdates, CreateResponseOptions options, String conversationId, String resumeResponseId, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.OpenAIResponsesChatClient.FromOpenAIStreamingResponseUpdatesAsync(IAsyncEnumerable`1 streamingResponseUpdates, CreateResponseOptions options, String conversationId, String resumeResponseId, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Extensions.AI.ConfigureOptionsChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.ConfigureOptionsChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.ConfigureOptionsChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Extensions.AI.FunctionInvokingChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.FunctionInvokingChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Extensions.AI.FunctionInvokingChatClient.GetStreamingResponseAsync(IEnumerable`1 messages, ChatOptions options, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.ChatClientAgent.RunCoreStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.ChatClientAgent.RunCoreStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.ChatClientAgent.RunCoreStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.ChatClientAgent.RunCoreStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.AIAgent.RunStreamingAsync(IEnumerable`1 messages, AgentSession session, AgentRunOptions options, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.AgentResponseExtensions.AsChatResponseUpdatesAsync(IAsyncEnumerable`1 responseUpdates)+MoveNext()
   at Microsoft.Agents.AI.AgentResponseExtensions.AsChatResponseUpdatesAsync(IAsyncEnumerable`1 responseUpdates)+MoveNext()
   at Microsoft.Agents.AI.AgentResponseExtensions.AsChatResponseUpdatesAsync(IAsyncEnumerable`1 responseUpdates)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIChatResponseUpdateStreamExtensions.FilterServerToolsFromMixedToolInvocationsAsync(IAsyncEnumerable`1 updates, List`1 clientTools, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIChatResponseUpdateStreamExtensions.FilterServerToolsFromMixedToolInvocationsAsync(IAsyncEnumerable`1 updates, List`1 clientTools, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIChatResponseUpdateStreamExtensions.FilterServerToolsFromMixedToolInvocationsAsync(IAsyncEnumerable`1 updates, List`1 clientTools, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.Shared.ChatResponseUpdateAGUIExtensions.AsAGUIEventStreamAsync(IAsyncEnumerable`1 updates, String threadId, String runId, JsonSerializerOptions jsonSerializerOptions, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.Shared.ChatResponseUpdateAGUIExtensions.AsAGUIEventStreamAsync(IAsyncEnumerable`1 updates, String threadId, String runId, JsonSerializerOptions jsonSerializerOptions, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.Shared.ChatResponseUpdateAGUIExtensions.AsAGUIEventStreamAsync(IAsyncEnumerable`1 updates, String threadId, String runId, JsonSerializerOptions jsonSerializerOptions, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIEndpointRouteBuilderExtensions.SaveSessionAfterStreamingAsync(IAsyncEnumerable`1 events, AIHostAgent hostAgent, String threadId, AgentSession session, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIEndpointRouteBuilderExtensions.SaveSessionAfterStreamingAsync(IAsyncEnumerable`1 events, AIHostAgent hostAgent, String threadId, AgentSession session, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIEndpointRouteBuilderExtensions.SaveSessionAfterStreamingAsync(IAsyncEnumerable`1 events, AIHostAgent hostAgent, String threadId, AgentSession session, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIServerSentEventsResult.WrapEventsAsSseItemsAsync(IAsyncEnumerable`1 events, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIServerSentEventsResult.WrapEventsAsSseItemsAsync(IAsyncEnumerable`1 events, CancellationToken cancellationToken)+MoveNext()
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIServerSentEventsResult.WrapEventsAsSseItemsAsync(IAsyncEnumerable`1 events, CancellationToken cancellationToken)+System.Threading.Tasks.Sources.IValueTaskSource<System.Boolean>.GetResult()
   at System.Net.ServerSentEvents.SseFormatter.WriteAsyncCore[T](IAsyncEnumerable`1 source, Stream destination, Action`2 itemFormatter, CancellationToken cancellationToken)
   at System.Net.ServerSentEvents.SseFormatter.WriteAsyncCore[T](IAsyncEnumerable`1 source, Stream destination, Action`2 itemFormatter, CancellationToken cancellationToken)
   at Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.AGUIServerSentEventsResult.ExecuteAsync(HttpContext httpContext)

Package Versions

Microsoft.Agents.AI.Hosting.AGUI.AspNetCore: 1.11.1-preview.260625.1, Azure.AI.Projects: 2.0.1, Microsoft.Agents.AI: 1.12.0, Microsoft.Agents.AI.Abstractions: 1.12.0, Microsoft.Agents.AI.Foundry: 1.5.0

.NET Version

.NET 10

Additional Context

I am using Tanstack AI on the client and it includes the "role": "reasoning" in the Messages.

I debug Microsoft.Agents.AI.Hosting.AGUI and it change the role from reasoning to assistant. Not sure if this will help.

[github-actions]

🤖 Automated triage reproduction notes (agent-authored — trust but verify)

Agent analysis

Repro: AGUIChatMessageExtensions.AsChatMessages() in dotnet/src/Microsoft.Agents.AI.AGUI/Shared/AGUIChatMessageExtensions.cs (lines 79-102) converts AGUIReasoningMessage without encryptedValue into ChatMessage(Assistant, [TextReasoningContent(text) { ProtectedData = null }]), which is then forwarded unchanged through ChatClientAgent to the OpenAI Responses API causing invalid_payload. Trigger: TanStack AI client includes a role=reasoning message without encryptedValue in the conversation payload. Minimal repro: send an AG-UI RunAgentInput with messages array containing {role: "reasoning", content: "...", id: "..."} (no encryptedValue) to an AGUI endpoint backed by a ChatClientAgent using OpenAIResponsesChatClient.

• Failing test: dotnet/tests/Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.IntegrationTests/ReasoningMessageTests.cs
• Files examined: dotnet/src/Microsoft.Agents.AI.AGUI/Shared/AGUIChatMessageExtensions.cs, dotnet/src/Microsoft.Agents.AI.AGUI/Shared/AGUIReasoningMessage.cs, dotnet/src/Microsoft.Agents.AI.AGUI/Shared/AGUIRoles.cs, dotnet/src/Microsoft.Agents.AI.AGUI/Shared/AGUIMessageJsonConverter.cs, dotnet/src/Microsoft.Agents.AI.Hosting.AGUI.AspNetCore/AGUIEndpointRouteBuilderExtensions.cs, dotnet/src/Microsoft.Agents.AI/ChatClient/ChatClientAgent.cs
• Tests run: dotnet/tests/Microsoft.Agents.AI.Hosting.AGUI.AspNetCore.IntegrationTests/ReasoningMessageTests.cs
• Reported version: 1.11.1-preview.260625.1
• Current version: 1.13.0-preview