Added intermediate responses and response streaming via a queue. Improved dialog logging.

client/interface.py

@@ -2,6 +2,7 @@
 
 import json
 import logging
+import queue
 from anthropic.types import text_block
 from gradio.components.chatbot import ChatMessage
 
@@ -15,6 +16,7 @@ dialog = gradio_funcs.get_dialog_logger(clear = True)
 
 async def agent_input(
         bridge: AnthropicBridge,
+        output_queue: queue.Queue,
         chat_history: list
 ) -> list:
 
@@ -46,6 +48,7 @@ async def agent_input(
             else:
                 intermediate_reply = f'I Will check the {website} RSS feed for you'
 
+            output_queue.put(intermediate_reply)
             dialog.info('LLM: %s', intermediate_reply)
             dialog.info('LLM: called %s on %s', tool_name, website)
 
@@ -63,6 +66,9 @@ async def agent_input(
                 'content': prompt
             }]
 
+            dialog.info('System: re-prompting LLM with return from %s call', tool_name)
+            dialog.info('New prompt: %s ...', prompt[:150])
+
             logger.info('Re-prompting input %s', input_message)
             result = await bridge.process_query(
                 prompts.GET_FEED_SYSTEM_PROMPT,
@@ -71,22 +77,12 @@ async def agent_input(
 
             try:
 
-                final_reply = result['llm_response'].content[0].text
+                reply = result['llm_response'].content[0].text
 
             except (IndexError, AttributeError):
-                final_reply = 'No final reply from model'
-
-            logger.info('LLM final reply: %s', final_reply)
+                reply = 'No final reply from model'
 
-            chat_history.append({
-                "role": "assistant",
-                "content": intermediate_reply
-            })
-
-            chat_history.append({
-                "role": "assistant",
-                "content": final_reply
-            })
+            logger.info('LLM final reply: %s', reply)
 
     else:
         try:
@@ -97,12 +93,9 @@ async def agent_input(
 
         logger.info('Direct, no-tool reply: %s', reply)
 
-        chat_history.append({
-            "role": "assistant",
-            "content": reply
-        })
-
-    return chat_history
+    dialog.info('LLM: %s ...', reply[:100])
+    output_queue.put(reply)
+    output_queue.put('bot-finished')
 
 
 def format_chat_history(history) -> list[dict]:
@@ -114,9 +107,9 @@ def format_chat_history(history) -> list[dict]:
         if isinstance(chat_message, ChatMessage):
             role, content = chat_message.role, chat_message.content
         else:
-            role, content = chat_message.get("role"), chat_message.get("content")
+            role, content = chat_message.get('role'), chat_message.get('content')
 
-        if role in ["user", "assistant", "system"]:
-            messages.append({"role": role, "content": content})
+        if role in ['user', 'assistant', 'system']:
+            messages.append({'role': role, 'content': content})
 
     return messages

client/prompts.py

@@ -5,19 +5,10 @@ from string import Template
 
 DEFAULT_SYSTEM_PROMPT = 'You are a helpful tool-using assistant.'
 
-GET_FEED_SYSTEM_PROMPT = '''
-You are a helpful assistant. Your job is to facilitate interactions between
-Human users and LLM agents.
-'''
+GET_FEED_SYSTEM_PROMPT = '''You are a helpful assistant. Your job is to facilitate interactions between Human users and LLM agents.'''
 
 GET_FEED_PROMPT = Template(
-'''
-Below is an exchange between a user and an agent. The user has asked 
-the agent to get new content from the $website RSS feed. In order to
-complete the request, the agent has called a function which returned
-the RSS feed content from $website in JSON format. Your job is to 
-complete the exchange by using the returned JSON RSS feed data to write
-a human readable reply to the user.
+'''Below is an exchange between a user and an agent. The user has asked the agent to get new content from the $website RSS feed. In order to complete the request, the agent has called a function which returned the RSS feed content from $website in JSON format. Your job is to complete the exchange by using the returned JSON RSS feed data to write a human readable reply to the user.
 
 user: $user_query
 
@@ -27,6 +18,5 @@ function call: get_feed_content($website)
 
 function return: $articles
 
-assistant:
-'''                       
-)
+assistant:'''
+)

rss_client.py

@@ -1,7 +1,11 @@
 '''RSS MCP server demonstration client app.'''
 
 import os
+import asyncio
 import logging
+import time
+import queue
+from typing import Tuple
 from pathlib import Path
 from logging.handlers import RotatingFileHandler
 import gradio as gr
@@ -48,23 +52,51 @@ BRIDGE = AnthropicBridge(
 
 logger.info('Started Anthropic API bridge')
 
+# Queue to return responses to user
+OUTPUT_QUEUE = queue.Queue()
+logger.info('Created response queue')
+
+def user_message(message: str, history: list) -> Tuple[str, list]:
+    '''Adds user message to conversation and returns for immediate posting.
 
-async def send_message(message: str, chat_history: list) -> str:
-    '''Submits user message to agent.
-    
     Args:
         message: the new message from the user as a string
         chat_history: list containing conversation history where each element is
             a dictionary with keys 'role' and 'content'
 
+    Returns
+        New chat history with user's message added.
+    '''
+
+    return '', history + [{'role': 'user', 'content': message}]
+
+
+def send_message(chat_history: list):
+    '''Submits chat history to agent, streams reply, one character at a time.
+    
+    Args:
+        chat_history: list containing conversation history where each element is
+            a dictionary with keys 'role' and 'content'
+
     Returns
         New chat history with model's response to user added.
     '''
 
-    chat_history.append({'role': 'user', 'content': message})
-    chat_history = await interface.agent_input(BRIDGE, chat_history)
+    asyncio.run(interface.agent_input(BRIDGE, OUTPUT_QUEUE, chat_history))
+
+    while True:
+        response = OUTPUT_QUEUE.get()
+
+        if response == 'bot-finished':
+            break
+
+        chat_history.append({'role': 'assistant', 'content': ''})
+
+        for character in response:
+            chat_history[-1]['content'] += character
+            time.sleep(0.005)
 
-    return '', chat_history
+            yield chat_history
 
 
 with gr.Blocks(title='MCP RSS client') as demo:
@@ -82,8 +114,8 @@ with gr.Blocks(title='MCP RSS client') as demo:
     )
 
     # Dialog log output
-    dialog_output = gr.Textbox(label='Internal dialog', lines=10, max_lines=10)
-    timer = gr.Timer(1, active=True)
+    dialog_output = gr.Textbox(label='Internal dialog', lines=10, max_lines=100)
+    timer = gr.Timer(0.5, active=True)
 
     timer.tick( # pylint: disable=no-member
         lambda: gradio_funcs.update_dialog(), # pylint: disable=unnecessary-lambda
@@ -106,9 +138,9 @@ with gr.Blocks(title='MCP RSS client') as demo:
     )
 
     msg.submit( # pylint: disable=no-member
-        send_message,
-        [msg, chatbot],
-        [msg, chatbot]
+        user_message, [msg, chatbot], [msg, chatbot], queue=False
+    ).then(
+        send_message, chatbot, chatbot
     )