Dynamically generate help message for slash commands in chat UI (#520)

* Show help for custom slash commands

* Morph `short_help` into existing `help` per review

packages/jupyter-ai/jupyter_ai/chat_handlers/ask.py

@@ -28,7 +28,7 @@ class AskChatHandler(BaseChatHandler):
 
     id = "ask"
     name = "Ask with Local Data"
-    help = "Asks a question with retrieval augmented generation (RAG)"
+    help = "Ask a question about your learned data"
     routing_type = SlashCommandRoutingType(slash_id="ask")
 
     def __init__(self, retriever, *args, **kwargs):

packages/jupyter-ai/jupyter_ai/chat_handlers/clear.py

@@ -8,7 +8,7 @@
 class ClearChatHandler(BaseChatHandler):
     id = "clear"
     name = "Clear chat messages"
-    help = "Clears the displayed chat message history only; does not clear the context sent to chat providers"
+    help = "Clear the chat window"
     routing_type = SlashCommandRoutingType(slash_id="clear")
 
     def __init__(self, *args, **kwargs):

packages/jupyter-ai/jupyter_ai/chat_handlers/generate.py

@@ -218,7 +218,7 @@ def create_notebook(outline):
 class GenerateChatHandler(BaseChatHandler):
     id = "generate"
     name = "Generate Notebook"
-    help = "Generates a Jupyter notebook, including name, outline, and section contents"
+    help = "Generate a Jupyter notebook from a text prompt"
     routing_type = SlashCommandRoutingType(slash_id="generate")
 
     def __init__(self, preferred_dir: str, log_dir: Optional[str], *args, **kwargs):

packages/jupyter-ai/jupyter_ai/chat_handlers/help.py

@@ -1,5 +1,5 @@
 import time
-from typing import List
+from typing import Dict
 from uuid import uuid4
 
 from jupyter_ai.models import AgentChatMessage, HumanChatMessage
@@ -8,34 +8,42 @@
 
 HELP_MESSAGE = """Hi there! I'm Jupyternaut, your programming assistant.
 You can ask me a question using the text box below. You can also use these commands:
-* `/learn` — Teach Jupyternaut about files on your system
-* `/ask` — Ask a question about your learned data
-* `/generate` — Generate a Jupyter notebook from a text prompt
-* `/clear` — Clear the chat window
-* `/help` — Display this help message
+{commands}
 
 Jupyter AI includes [magic commands](https://jupyter-ai.readthedocs.io/en/latest/users/index.html#the-ai-and-ai-magic-commands) that you can use in your notebooks.
 For more information, see the [documentation](https://jupyter-ai.readthedocs.io).
 """
 
 
-def HelpMessage():
+def _format_help_message(chat_handlers: Dict[str, BaseChatHandler]):
+    commands = "\n".join(
+        [
+            f"* `{command_name}` — {handler.help}"
+            for command_name, handler in chat_handlers.items()
+            if command_name != "default"
+        ]
+    )
+    return HELP_MESSAGE.format(commands=commands)
+
+
+def HelpMessage(chat_handlers: Dict[str, BaseChatHandler]):
     return AgentChatMessage(
         id=uuid4().hex,
         time=time.time(),
-        body=HELP_MESSAGE,
+        body=_format_help_message(chat_handlers),
         reply_to="",
     )
 
 
 class HelpChatHandler(BaseChatHandler):
     id = "help"
     name = "Help"
-    help = "Displays a help message in the chat message area"
+    help = "Display this help message"
     routing_type = SlashCommandRoutingType(slash_id="help")
 
-    def __init__(self, *args, **kwargs):
+    def __init__(self, *args, chat_handlers: Dict[str, BaseChatHandler], **kwargs):
         super().__init__(*args, **kwargs)
+        self._chat_handlers = chat_handlers
 
     async def process_message(self, message: HumanChatMessage):
-        self.reply(HELP_MESSAGE, message)
+        self.reply(_format_help_message(self._chat_handlers), message)

packages/jupyter-ai/jupyter_ai/chat_handlers/learn.py

@@ -33,7 +33,7 @@
 class LearnChatHandler(BaseChatHandler):
     id = "learn"
     name = "Learn Local Data"
-    help = "Pass a list of files and directories. Once converted to vector format, you can ask about them with /ask."
+    help = "Teach Jupyternaut about files on your system"
     routing_type = SlashCommandRoutingType(slash_id="learn")
 
     def __init__(self, *args, **kwargs):

packages/jupyter-ai/jupyter_ai/extension.py

@@ -153,7 +153,7 @@ def initialize_settings(self):
         # list of chat messages to broadcast to new clients
         # this is only used to render the UI, and is not the conversational
         # memory object used by the LM chain.
-        self.settings["chat_history"] = [HelpMessage()]
+        self.settings["chat_history"] = []
 
         # get reference to event loop
         # `asyncio.get_event_loop()` is deprecated in Python 3.11+, in favor of
@@ -189,7 +189,6 @@ def initialize_settings(self):
             log_dir=self.error_logs_dir,
         )
         learn_chat_handler = LearnChatHandler(**chat_handler_kwargs)
-        help_chat_handler = HelpChatHandler(**chat_handler_kwargs)
         retriever = Retriever(learn_chat_handler=learn_chat_handler)
         ask_chat_handler = AskChatHandler(**chat_handler_kwargs, retriever=retriever)
 
@@ -199,9 +198,12 @@ def initialize_settings(self):
             "/clear": clear_chat_handler,
             "/generate": generate_chat_handler,
             "/learn": learn_chat_handler,
-            "/help": help_chat_handler,
         }
 
+        help_chat_handler = HelpChatHandler(
+            **chat_handler_kwargs, chat_handlers=jai_chat_handlers
+        )
+
         slash_command_pattern = r"^[a-zA-Z0-9_]+$"
         for chat_handler_ep in chat_handler_eps:
             try:
@@ -248,6 +250,12 @@ def initialize_settings(self):
                 f"Registered chat handler `{chat_handler.id}` with command `{command_name}`."
             )
 
+        # Make help always appear as the last command
+        jai_chat_handlers["/help"] = help_chat_handler
+
+        self.settings["chat_history"].append(
+            HelpMessage(chat_handlers=jai_chat_handlers)
+        )
         self.settings["jai_chat_handlers"] = jai_chat_handlers
 
         latency_ms = round((time.time() - start) * 1000)