Custom Action registered, but not being called

Rasa Pro CALM
1

I’ve started creating a Custom Action for managing survey ratings, and I can see that the custom_class.py is getting registered in the action server, but it never seems to get called. I think my slot definition points to it, but it’s acting like it never decides to call.

class SetRating(Action):

    def name(self) -> str:
        print("set_rating.name ENTERED")
        return "set_rating"

    def run(self, dispatcher: CollectingDispatcher,
            tracker: Tracker, domain: Dict[str, Any]):

        print("set_rating ENTERED")
        print("dispatcher:", dispatcher)
        print("tracker:", tracker)
        print("domain:", domain)
        # account = get_account(tracker.sender_id)
        print("set_rating LEAVING")

        return [SlotSet("survey_question_applying_ai_in_organization", "3")]

and the slot is defined as:

slots:

  survey_question_understanding_genai:
    type: float
    min_value: 1
    max_value: 5
    initial_value: null
    influence_conversation: false
    mappings:
      - type: custom
        action: set_rating

actions:
  - set_rating

responses:
  utter_ask_survey_question_understanding_genai:
  - text:  "On a scale of 1 to 5, how much do YOU AGREE that Understanding GenAI is important to you in your work? (1=Strongly Disagree, 5=Strongly Agree)"

and my endpoints are configured:

action_endpoint:
  url: "http://localhost:5055/webhook"

The response after the utterance should call the custom action, right? I see the " ``` set_rating.name ENTERED

2

Hi John, is this in an NLU + LLM CALM bot, or LLM-only?

Can you share your flow where you’re attempting to gather the feedback?

3

Thanks Chris. I just noticed your response yesterday, but unfortunately have gotten rid of all of the code, and replaced it with “type: from_llm” code that doesn’t require custom gatherers. The stuff I was writing was all CALM bot, but I may have been building it ontop of the demo code that was goes both ways.

When I reported the original question, I was assuming that I could simply set my slots with

  survey_question_metrics_in_government:
    type: float
    influence_conversation: false
    mappings:
      - type: custom
        action: set_rating

and then when the “collect” for that slot was stepped to in the following flow:

flows:
  collect_survey_results:
    description: Get student feedback to a course survey.
    steps:
      - action: utter_survey_start
      - collect: survey_question_metrics_in_government
      - collect: survey_question_behavioral_insights
      - collect: survey_question_rate_general_knowledge
      - collect: survey_question_what_tools_techniques_buzzwords
      - collect: survey_question_how_you_use_gen_ai_now
      - action: utter_survey_complete

that it would automatically call set_rating.py. But it didn’t do that, and I didn’t get it working.

However, my from_llm version of the collect is doing what I need right now, so I’m probably OK.