MiddlewareChatClient<T>
Models & Types · AiDotNet.Agentic.Pipeline
An IChatClient decorator that runs a chain of IChatMiddleware around an
inner client’s calls — the composition root for filters/middleware (logging, guardrails, caching, retry,
telemetry). The first-registered middleware runs outermost.
Implements: IChatClientDecorator<T>
For Beginners
Wrap your model client in this with a list of filters, and every (non-streaming) call flows through them in order before and after hitting the model — one place to add behavior that should apply everywhere.
How It Works
Middleware semantics (pre/post processing, short-circuiting) are defined on the complete-response path.
When middleware are configured, GetStreamingResponseAsync therefore runs the full pipeline
and re-emits the final response as a stream — streaming callers get the exact same policy enforcement as
non-streaming callers, at the cost of token-level incrementality. With no middleware configured, streaming
passes through to the inner client natively.
Constructors
| Constructor | Summary |
|---|---|
MiddlewareChatClient(IChatClient<T> inner, IReadOnlyList<IChatMiddleware> middlewares) |
Initializes a new middleware client. |
Properties
| Property | Summary |
|---|---|
IChatClient<T> Inner { get; } |
|
string ModelId { get; } |
Methods
| Method | Summary |
|---|---|
IAsyncEnumerable<ChatResponseUpdate> GetStreamingResponseAsync(IReadOnlyList<ChatMessage> messages, ChatOptions options, CancellationToken cancellationToken) |
|
Task<ChatResponse> GetResponseAsync(IReadOnlyList<ChatMessage> messages, ChatOptions options, CancellationToken cancellationToken) |