SequentialAgent Authentication issue

[denamwangi]

Hi! I'm working through the auth documentation and I can successfully get the authentication flow to work with the single agent example. When I move to a multiagent scenario (SequentialAgent or Agent as orchestrator), my flow breaks. It's possible i'm missing some adjustment in the async events from the runner that needs tweaking for multiagent scenarios but i'm not sure (example code here). I get a Function auth_spotify is not found in the tools_dict error.

class SpotifySearchTags(BaseModel):
    title: str = Field(..., description="A fun title for the playlist ")
    mood: str = Field(..., description="1-3 space separated descriptors of the type of mood the playlist should have e.g. 'happy emo' ")
    genre: Optional[str] = Field(None, description="The musical genre e.g. afrobeats, pop, R&B")
    

get_music_tags_agent = Agent(
    name="get_music_tags_agent",
    model=LiteLlm(cheap_model),
    description="Generates a playlist title and musical tags based on the destination city which will be used to get tracks from spotify",
    instruction="You are a musical expert generating playlist tags based on a user's travel destination. "
                "Your output should include a mood and genre suitable for the city and time of year. "
                "Respond in structured JSON using the SpotifySearchTags schema and use only the fields included."
                " Include 2-3 for each field. For example for Nairobi as the destination you might return" 
                "SpotifySearchTags(title='Breezy Summer in Paris' mood='upbeat happy' genre='afrobeats amapiano')"
                "Your output must be a pure JSON object and must not include any text, commentary, or Markdown formatting. "
                "Return only the JSON object with fields 'mood' and 'genre'.",
    output_key="spotify_search_tags",
)

auth_spotify_agent = Agent(
    name="auth_spotify_agent",
    model=LiteLlm(orchestrator_model),
    description="Authenticated a user to spotify",
    instruction=(
        "Your job is to authenticate with Spotify "
        "Use the 'auth_spotify' tool to do this. "
        "You do not need to pass any input to the tool — just call it when appropriate. "
    ),                
    tools=[auth_spotify]
)


music_playlist_pipeline_agent = SequentialAgent(
    name="music_playlist_pipeline_agent",
    sub_agents=[get_music_tags_agent, auth_spotify_agent],
    description="Spotify playlist "
)```

and here's a snippet of the runner section (only a selection of it) which follows the example in auth:

async for event in events_async:
    print('%'*50)
    print(f" Event Info: Author: {event.author}, Type: {type(event).__name__}, Final: {event.is_final_response()} Content: {event.content}")
    print('%'*50)
    if event.is_final_response():
        if event.content and event.content.parts:
            # Assuming text response in the first part
            final_response_text = event.content.parts[0].text
        elif event.actions and event.actions.escalate: # Handle potential errors/escalations
            final_response_text = f"Agent escalated: {event.error_message or 'No specific message.'}"
        # Add more checks here if needed (e.g., specific error codes)
        print(f" <<< Agent Response: {final_response_text}")

    if is_pending_auth_event(event):
        auth_request_event_id = get_function_call_id(event)
        break

auth_content = types.Content(
    role='user',
    parts=[
        types.Part(
            function_response=types.FunctionResponse(
                id=auth_request_event_id,
                name='adk_request_credential',
                response=auth_config.model_dump()
            )
        )
    ]
)


events_async = runner.run_async(
  session_id=session_id,
  user_id=user_id,
  new_message=auth_content
)

async for event in events_async:
    print(f" Event Info: Author: {event.author}, Type: {type(event).__name__}, Final: {event.is_final_response()} Content: {event.content}")