more changes

crates/brightstaff/src/main.rs

@@ -142,8 +142,9 @@ async fn main() -> Result<(), Box<dyn std::error::Error + Send + Sync>> {
                         .with_context(parent_cx)
                         .await
                     }
-                    (&Method::GET, "/v1/models") => Ok(list_models(llm_providers).await),
-                    (&Method::OPTIONS, "/v1/models") => {
+                    (&Method::GET, "/v1/models" | "/agents/v1/models") => Ok(list_models(llm_providers).await),
+                    // hack for now to get openw-web-ui to work
+                    (&Method::OPTIONS, "/v1/models" | "/agents/v1/models") => {
                         let mut response = Response::new(empty());
                         *response.status_mut() = StatusCode::NO_CONTENT;
                         response

demos/use_cases/rag_agent/docker-compose.yaml

@@ -7,7 +7,7 @@ services:
       - "4317:4317"
       - "4318:4318"
   open-web-ui:
-    image: ghcr.io/open-webui/open-webui:main
+    image: dyrnq/open-webui:main
     restart: always
     ports:
       - "8080:8080"

demos/use_cases/rag_agent/src/rag_agent/response_generator_agent.py

@@ -75,6 +75,9 @@ async def chat_completions(request_body: ChatCompletionRequest, request: Request
         return StreamingResponse(
             stream_chat_completions(request_body, traceparent_header),
             media_type="text/plain",
+            headers={
+                "content-type": "text/event-stream",
+            },
         )
     else:
         return await non_streaming_chat_completions(request_body, traceparent_header)