Local Setup for Custom Actions - Rasa Documentation

This guide is for technical users who want a quick way to test out a new custom action from a local development server.

Prerequisites

This workflow requires that you already have a Rasa CALM project initialized. Please see the Rasa installation guide for more details.

Step 1: Implement Your Custom Action

Add your custom action by creating or modifying the actions/actions.py file. Below is an example of a custom action that validates sufficient funds:

actions.py

# This file contains your custom actions which can be used to run
# custom Python code.
# See Rasa Tutorial for more details https://rasa.com/docs/pro/tutorial#integrating-an-api-call

from typing import Any, Text, Dict, List
from rasa_sdk import Action, Tracker
from rasa_sdk.executor import CollectingDispatcher
from rasa_sdk.events import SlotSet

class ActionValidateSufficientFunds(Action):

    def name(self) -> Text:
        return "action_validate_sufficient_funds"

    def run(self, dispatcher: CollectingDispatcher,
            tracker: Tracker,
            domain: Dict[Text, Any]) -> List[Dict[Text, Any]]:
        # Hard-coded balance for tutorial purposes. In production,
        # this would be retrieved from a database or an API.
        balance = 1000
        transfer_amount = tracker.get_slot("amount")
        has_sufficient_funds = transfer_amount <= balance
        return [SlotSet("has_sufficient_funds", has_sufficient_funds)]

Step 2: Run the Action Server

Start the action server locally using the Rasa CLI:

rasa run actions

This command launches the action server, which runs on http://localhost:5055 by default.

Step 3: Expose the Action Server

For Rasa Studio to connect to your custom action server during testing, expose your local server to the public internet. You can use a tool like ngrok for this purpose. Run the following command:

ngrok http 5055

Optional: Deploy to a Cloud Environment

If you want to avoid using ngrok for repeated testing or need a more robust setup, see more details on the action server reference to deploy your development action server to a cloud environment of your choice. Ensure the public URL of your deployed action server is accessible and provide it to Rasa Studio for integration.

Step 4: Connect to Studio

Open Rasa Studio and load your assistant project.
Navigate to the Settings section in Rasa Studio.
Go to the Endpoints tab and locate the Action Server URL field.
Enter the URL of your action server:
- Option 1 (Using ngrok): Paste the public URL generated by the ngrok http 5055 command (e.g., https://<your-ngrok-subdomain>.ngrok.io).
- Option 2 (Local Server): If you are running the action server locally and testing on the same machine, use http://localhost:5055.
Click Save to apply the configuration.
Train your assistant by clicking the Train button to ensure the new custom action is included in the training process.
Use the Interactive Learning or Conversations tools to test your assistant and confirm that the custom action is executed as expected.

Troubleshooting Tips

If the assistant cannot connect to the action server:

Ensure the action server is running (rasa run actions).
If using ngrok, verify the public URL is still active and hasn’t expired.
Confirm the URL matches what is specified in the Action Server URL field.

​Prerequisites

​Step 1: Implement Your Custom Action

​Step 2: Run the Action Server

​Step 3: Expose the Action Server

​Optional: Deploy to a Cloud Environment

​Step 4: Connect to Studio

​Troubleshooting Tips