feat: feedback block

README.md

@@ -51,6 +51,8 @@ pip install -r requirements.txt
 python3 app.py
 ```
 
+Start talking to the bot! Start a new DM or thread and click the feedback button when it responds.
+
 #### Linting
 ```zsh
 # Run flake8 from root directory for linting
@@ -74,8 +76,8 @@ black .
 
 Every incoming request is routed to a "listener". This directory groups each listener based on the Slack Platform feature used, so `/listeners/events` handles incoming events, `/listeners/shortcuts` would handle incoming [Shortcuts](https://docs.slack.dev/interactivity/implementing-shortcuts/) requests, and so on.
 
-:::info[The `listeners/events` folder is purely educational and demonstrates alternative approaches to implementation] 
-These listeners are **not registered** and are not used in the actual application. For the working implementation, refer to `listeners/assistant.py`.
+> [!NOTE] 
+> The `listeners/events` folder is purely educational and demonstrates alternative approaches to implementation. These listeners are **not registered** and are not used in the actual application. For the working implementation, refer to `listeners/assistant/assistant.py`.
 
 **`/listeners/assistant`**
 

listeners/__init__.py

@@ -1,9 +1,11 @@
 from listeners.assistant import assistant
+from listeners.assistant.assistant import handle_feedback
 
 
 def register_listeners(app):
     # Using assistant middleware is the recommended way.
     app.assistant(assistant)
+    app.action("feedback")(handle_feedback)
 
     # The following event listeners demonstrate how to implement the same on your own.
     # from listeners import events

listeners/assistant/assistant.py

@@ -3,13 +3,46 @@
 from slack_bolt import Assistant, BoltContext, Say, SetSuggestedPrompts
 from slack_bolt.context.get_thread_context import GetThreadContext
 from slack_sdk import WebClient
+from slack_sdk.models.blocks import FeedbackButtonsElement, FeedbackButtonObject, ContextActionsBlock
 
 from ..llm_caller import call_llm
 
 # Refer to https://tools.slack.dev/bolt-python/concepts/assistant/ for more details
 assistant = Assistant()
 
 
+def create_feedback_block(user_id: str) -> ContextActionsBlock:
+    """
+    Create feedback block with thumbs up/down buttons
+
+    Args:
+        user_id: User ID for user-specific controls
+
+    Returns:
+        Block Kit context_actions block
+    """
+    block = [
+        ContextActionsBlock(
+            elements=[
+                FeedbackButtonsElement(
+                    action_id="feedback",
+                    positive_button=FeedbackButtonObject(
+                        text="Good Response",
+                        accessibility_label="Submit positive feedback on this response",
+                        value="good-feedback",
+                    ),
+                    negative_button=FeedbackButtonObject(
+                        text="Bad Response",
+                        accessibility_label="Submit negative feedback on this response",
+                        value="bad-feedback",
+                    ),
+                )
+            ]
+        )
+    ]
+    return block
+
+
 # This listener is invoked when a human user opened an assistant thread
 @assistant.thread_started
 def start_assistant_thread(
@@ -60,6 +93,7 @@ def respond_in_assistant_thread(
     say: Say,
 ):
     try:
+        user_id = payload["user"]
         channel_id = payload["channel"]
         thread_ts = payload["thread_ts"]
 
@@ -83,26 +117,56 @@ def respond_in_assistant_thread(
             messages_in_thread.append({"role": role, "content": message["text"]})
 
         returned_message = call_llm(messages_in_thread)
+
         client.assistant_threads_setStatus(
             channel_id=channel_id, thread_ts=thread_ts, status="Bolt is typing", loading_messages=loading_messages
         )
+
         stream_response = client.chat_startStream(
             channel=channel_id,
             thread_ts=thread_ts,
         )
         stream_ts = stream_response["ts"]
+
         # use of this for loop is specific to openai response method
         for event in returned_message:
             if event.type == "response.output_text.delta":
                 client.chat_appendStream(channel=channel_id, ts=stream_ts, markdown_text=f"{event.delta}")
             else:
                 continue
 
-        client.chat_stopStream(
-            channel=channel_id,
-            ts=stream_ts,
-        )
+        feedback_block = create_feedback_block(user_id=user_id)
+        client.chat_stopStream(channel=channel_id, ts=stream_ts, blocks=feedback_block)
 
     except Exception as e:
         logger.exception(f"Failed to handle a user message event: {e}")
         say(f":warning: Something went wrong! ({e})")
+
+
+# Handle feedback buttons (thumbs up/down)
+def handle_feedback(ack, body, client, logger):
+    ack()
+    try:
+        message_ts = body["message"]["ts"]
+        channel_id = body["channel"]["id"]
+        feedback_type = body["actions"][0]["value"]
+        is_positive = feedback_type == "good-feedback"
+
+        if is_positive:
+            client.chat_postEphemeral(
+                channel=channel_id,
+                user=body["user"]["id"],
+                thread_ts=message_ts,
+                text="We're glad you found this useful.",
+            )
+        else:
+            client.chat_postEphemeral(
+                channel=channel_id,
+                user=body["user"]["id"],
+                thread_ts=message_ts,
+                text="Sorry to hear that response wasn't up to par :slightly_frowning_face: Starting a new chat may help with AI mistakes and hallucinations.",
+            )
+
+        logger.debug(f"Handled feedback: type={feedback_type}, message_ts={message_ts}")
+    except Exception as error:
+        logger.error(f":warning: Something went wrong! {error}")