> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/19.latest/extending/middleware/chat-middleware.md).

# Chat Middleware

Create middleware to intercept and modify chat operations.

Chat middleware wraps `IChatClient` instances to add behavior to chat completions. Implement `IAIChatMiddleware` to create custom middleware.

## Interface

{% code title="IAIChatMiddleware" %}

```csharp
public interface IAIChatMiddleware
{
    IChatClient Apply(IChatClient client);
}
```

{% endcode %}

The `Apply` method receives the current client and returns a wrapped client.

## Basic Implementation

{% code title="SimpleChatMiddleware.cs" %}

```csharp
using Microsoft.Extensions.AI;
using Umbraco.AI.Core.Chat;

public class SimpleChatMiddleware : IAIChatMiddleware
{
    public IChatClient Apply(IChatClient client)
    {
        return new SimpleChatClientWrapper(client);
    }
}
```

{% endcode %}

## Creating a Custom Wrapper

To implement custom behavior, create a class that delegates to the inner client. The recommended approach is to derive from `DelegatingChatClient` (provided by `Microsoft.Extensions.AI`), which handles pass-through for any members you don't override:

{% code title="MetricsChatMiddleware.cs" %}

```csharp
using System.Diagnostics;
using System.Runtime.CompilerServices;
using Microsoft.Extensions.AI;
using Microsoft.Extensions.Logging;
using Umbraco.AI.Core.Chat;

public class MetricsChatMiddleware : IAIChatMiddleware
{
    private readonly ILogger<MetricsChatMiddleware> _logger;

    public MetricsChatMiddleware(ILogger<MetricsChatMiddleware> logger)
    {
        _logger = logger;
    }

    public IChatClient Apply(IChatClient client)
    {
        return new MetricsChatClient(client, _logger);
    }
}

internal sealed class MetricsChatClient : DelegatingChatClient
{
    private readonly ILogger _logger;

    public MetricsChatClient(IChatClient innerClient, ILogger logger)
        : base(innerClient)
    {
        _logger = logger;
    }

    public override async Task<ChatResponse> GetResponseAsync(
        IEnumerable<ChatMessage> messages,
        ChatOptions? options = null,
        CancellationToken cancellationToken = default)
    {
        var stopwatch = Stopwatch.StartNew();

        try
        {
            var response = await base.GetResponseAsync(messages, options, cancellationToken);
            stopwatch.Stop();

            _logger.LogInformation(
                "Chat completed in {ElapsedMs}ms. Tokens: {Input}/{Output}",
                stopwatch.ElapsedMilliseconds,
                response.Usage?.InputTokenCount,
                response.Usage?.OutputTokenCount);

            return response;
        }
        catch (Exception ex)
        {
            stopwatch.Stop();
            _logger.LogError(ex, "Chat failed after {ElapsedMs}ms", stopwatch.ElapsedMilliseconds);
            throw;
        }
    }

    public override async IAsyncEnumerable<ChatResponseUpdate> GetStreamingResponseAsync(
        IEnumerable<ChatMessage> messages,
        ChatOptions? options = null,
        [EnumeratorCancellation] CancellationToken cancellationToken = default)
    {
        var stopwatch = Stopwatch.StartNew();
        var updateCount = 0;

        await foreach (var update in base.GetStreamingResponseAsync(messages, options, cancellationToken))
        {
            updateCount++;
            yield return update;
        }

        stopwatch.Stop();
        _logger.LogInformation(
            "Streaming completed in {ElapsedMs}ms. Updates: {Count}",
            stopwatch.ElapsedMilliseconds,
            updateCount);
    }
}
```

{% endcode %}

## Using M.E.AI Built-in Middleware

Microsoft.Extensions.AI (M.E.AI) provides built-in middleware options. Use the builder pattern:

{% code title="Using Built-in Middleware" %}

```csharp
using Microsoft.Extensions.AI;
using Microsoft.Extensions.Logging;

public class LoggingChatMiddleware : IAIChatMiddleware
{
    private readonly ILoggerFactory _loggerFactory;

    public LoggingChatMiddleware(ILoggerFactory loggerFactory)
    {
        _loggerFactory = loggerFactory;
    }

    public IChatClient Apply(IChatClient client)
    {
        return client.AsBuilder()
            .UseLogging(_loggerFactory)
            .Build();
    }
}
```

{% endcode %}

### Available M.E.AI Middleware

| Method                    | Purpose                      |
| ------------------------- | ---------------------------- |
| `UseLogging()`            | Log requests and responses   |
| `UseOpenTelemetry()`      | Add OpenTelemetry tracing    |
| `UseFunctionInvocation()` | Enable tool/function calling |
| `UseDistributedCache()`   | Cache responses              |

## Registering Chat Middleware

Register middleware in a Composer:

{% code title="MyComposer.cs" %}

```csharp
using Umbraco.AI.Extensions;
using Umbraco.Cms.Core.Composing;

public class MyComposer : IComposer
{
    public void Compose(IUmbracoBuilder builder)
    {
        builder.AIChatMiddleware()
            .Append<MetricsChatMiddleware>();
    }
}
```

{% endcode %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.umbraco.com/ai-in-umbraco/19.latest/extending/middleware/chat-middleware.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.