Module interceptor 

Interceptor system for middleware and observability.

Interceptors provide hooks into the request/response lifecycle, enabling:

• Telemetry and tracing
• Logging and debugging
• Metrics collection
• Request/response transformation
• Custom error handling
• Authentication enhancement

Architecture

The interceptor system follows a chain-of-responsibility pattern where multiple interceptors can be registered and executed in order. Each interceptor can:

• Modify request context before sending
• Observe and react to responses
• Handle streaming chunks
• Process errors

Example

use openai_ergonomic::{Client, Interceptor, BeforeRequestContext};

struct LoggingInterceptor;

#[async_trait::async_trait]
impl Interceptor for LoggingInterceptor {
    async fn before_request(&self, ctx: &mut BeforeRequestContext<'_>) -> Result<()> {
        println!("Calling {} with model {}", ctx.operation, ctx.model);
        Ok(())
    }
}

let client = Client::from_env()?
    .with_interceptor(Box::new(LoggingInterceptor))
    .build();

Structs

AfterResponseContext

Context provided after a successful non-streaming response.

BeforeRequestContext

Context provided before a request is sent.

ErrorContext

Context provided when an error occurs.

InterceptorChain

A chain of interceptors that are executed in order.

StreamChunkContext

Context provided for each chunk in a streaming response.

StreamEndContext

Context provided when a streaming response completes.

Traits

Interceptor

Trait for implementing interceptors.