add option to disable topic summary

Author: Radovan Fuchs

src/app/endpoints/query.py

@@ -312,9 +312,17 @@ async def query_endpoint_handler_base(  # pylint: disable=R0914
                 session.query(UserConversation).filter_by(id=conversation_id).first()
             )
             if not existing_conversation:
-                topic_summary = await get_topic_summary_func(
-                    query_request.query, client, llama_stack_model_id
-                )
+                # Check if topic summary should be generated (default: True)
+                should_generate = query_request.generate_topic_summary
+                
+                if should_generate:
+                    logger.debug("Generating topic summary for new conversation")
+                    topic_summary = await get_topic_summary_func(
+                        query_request.query, client, llama_stack_model_id
+                    )
+                else:
+                    logger.debug("Topic summary generation disabled by request parameter")
+                    topic_summary = None
         # Convert RAG chunks to dictionary format once for reuse
         logger.info("Processing RAG chunks...")
         rag_chunks_dict = [chunk.model_dump() for chunk in summary.rag_chunks]

src/app/endpoints/streaming_query.py

@@ -862,9 +862,17 @@ async def response_generator(
                     .first()
                 )
                 if not existing_conversation:
-                    topic_summary = await get_topic_summary(
-                        query_request.query, client, model_id
-                    )
+                    # Check if topic summary should be generated (default: True)
+                    should_generate = query_request.generate_topic_summary
+                    
+                    if should_generate:
+                        logger.debug("Generating topic summary for new conversation")
+                        topic_summary = await get_topic_summary(
+                            query_request.query, client, model_id
+                        )
+                    else:
+                        logger.debug("Topic summary generation disabled by request parameter")
+                        topic_summary = None
 
             completed_at = datetime.now(UTC).strftime("%Y-%m-%dT%H:%M:%SZ")
 

src/models/requests.py

@@ -81,6 +81,7 @@ class QueryRequest(BaseModel):
         system_prompt: The optional system prompt.
         attachments: The optional attachments.
         no_tools: Whether to bypass all tools and MCP servers (default: False).
+        generate_topic_summary: Whether to generate topic summary for new conversations (default: None = use server config).
         media_type: The optional media type for response format (application/json or text/plain).
 
     Example:
@@ -146,6 +147,12 @@ class QueryRequest(BaseModel):
         examples=[True, False],
     )
 
+    generate_topic_summary: Optional[bool] = Field(
+        True,
+        description="Whether to generate topic summary for new conversations (None = use server config default)",
+        examples=[True, False],
+    )
+
     media_type: Optional[str] = Field(
         None,
         description="Media type for the response format",
@@ -164,6 +171,7 @@ class QueryRequest(BaseModel):
                     "model": "model-name",
                     "system_prompt": "You are a helpful assistant",
                     "no_tools": False,
+                    "generate_topic_summary": True,
                     "attachments": [
                         {
                             "attachment_type": "log",