- 07 Jun 2024
- 6 Minutes to read
- Print
- PDF
Actions
- Updated on 07 Jun 2024
- 6 Minutes to read
- Print
- PDF
Actions are server-side functions executed by the chatbot as part of a conversational flow. Actions have the power to:
- Alter the state of the conversation.
- Send customized messages to the conversation.
- Execute arbitrary code like calling an API or storing data in the database.
Since they are JavaScript functions, they can do pretty much anything and have the following properties:
- user: all user attributes.
- session: variables kept only for the session.
- temp: variables kept only for the flow.
- bot: object containing global variables for this bot (same for all users).
- event: original (latest) event received from the user.
- args: arguments passed to this action from the Flow editor.
- process: sandboxed VM containing some of the env-variables (starting with EXPOSED_).
Registering New Actions
There are two ways to register new actions:
- You can add your JavaScript code in a .js file and put it in the folder data/global/actions.Note:There is no way to add new ones during runtime programmatically.
- You can also write actions directly in the Conversation Studio GUI by navigating to the code editor and using an Action Template.
We use JavaDoc comments to display meaningful information (such as name, description, arguments, default values) on the dialog flow editor. It is possible to keep an action hidden in the flow editor by adding the flag @hidden true in the JavaDoc.
External Libraries
Code in actions should be kept relatively simple. You can set counters, format data, fetch data from an API using the included axios library. If you need to use NPM modules/libraries or libraries from other programming languages, we recommend you either set up a server for that purpose or use serverless functions and then call the API using Digital Agent Actions. Using Digital Agent for heavy JavaScript computation use cases like image/video processing may cause latency issues for your bot.
Disabling Actions
Digital Agent will ignore files starting with a dot (.). This way you can disable a Hook or Action by merely prefixing the file's name with a dot.
Built-In Actions
Wait Action
- Action Title: Wait/Delay
- Category: Utility
- Author: Digital Agent, Inc.
- Parameters: data_type:number name:delay (default = 1000) - The number of milliseconds to wait
Example:
const wait = async delay => { return new Promise(resolve => setTimeout(() => resolve(), delay)) } return wait(args.delay || 1000 ) |
Append Context
- Action Title: Append Context
- Category: NLU
- Author: Digital Agent, Inc.
- 1st Parameter: data_type:string name:contexts - Comma-separated list of contexts
- 2nd Parameter: data_type:string name:[ttl=1] - Time-To-Live of the context in number of dialog turns. Put 0 to disable expiry.
This action adds context(s) to the list of contexts used by the NLU Engine for subsequent messages for that chat session.
- If a context were already present in the list, the higher TTL would win.
- To override a specific context, use the removeContext action before this action.
This method is contextual to the current user chat session. You can specify more than one context by separating them with a comma.
Remove Context
- Action Title: Remove Context
- Category: NLU
- Author: Digital Agent, Inc.
- Parameter: data_type:string name:contexts - Comma-separated list of contexts
Use this action to remove the provided context(s) from the list of contexts used by the NLU Engine for the subsequent messages for that chat session. This method is contextual to the current user chat session. You can specify more than one context by separating them with a comma.
Reset Context
- Action Title: Reset Context
- Category: NLU
- Author: Digital Agent, Inc.
- Parameter: none
It resets the NLU context to the default scope. This method is contextual to the current user and current chat session.
Send Feedback
- Action Title: Send Feedback
- Category: NDU66
- Author: Digital Agent, Inc.
- Parameter: data_type:number name:value - The feedback value. Use 1 for feedback, -1 for negative feedback
Provides feedback (1 for positive or -1 for negative feedback) at the end of a goal (a workflow that the user has completed).
Get Global Variable
- Action Title: Get global variable
- Category: Storage
- Author: Digital Agent, Inc.
- 1st Parameter: data_type:string name:name - The name of the variable
- 2nd Parameter: data_type:string name:output - The state variable to ouput
This action retrieves a variable that was stored globally using a storage key. Digital Agent uses a key: value storage system to allow complex object storage definitions.
Reset Global Variable
- Action Title: Reset Global Variable
- Category: Storage
- Author: Digital Agent, Inc.
- Parameter: data_type:string name:name - The name of the variable to be reset
Use this action to reset a variable with global scope.
Set Global Variable
- Action Title: Set global variable
- Category: Storage
- Author: Digital Agent, Inc.
- 1st Parameter: data_type:string name:name - The name of the variable
- 2nd Parameter: data_type:any name:value - Set the value of the variable
- 3rd Parameter: data_type:string name: [expiry=never] - Set the expiry of the data, can be never or a short string like 6 hours
- 4th Parameter: data_type:string name:output - The state variable to output to.
This action allows you to set a variable globally, with optional expiry.
Set Variable
- Action Title: Set Variable
- Category: Storage
- Author: Digital Agent, Inc.
- 1st Parameter: data_type:string name:type - Pick between user, session, temp, bot.
- 2nd Parameter: data_type:string name:name - The name of the variable.
- 3rd Parameter: data_type:any name:value - Set the value of the variable. Type null or leave empty to erase it.
You can use this Action to store data to desired storage based on the time to live expectation.
Reset Session
- Action Title: Reset Session
- Category: Storage
- Author: Digital Agent, Inc.
- Parameter: none
This action resets the user session and clears information stored in temp and session storage for the user. This action doesn't remove NLU Contexts and Last Messages history.
Switch Language
- Action Title: Switch Language
- Category: Language
- Author: Digital Agent, Inc.
- Parameter: data_type:string name:lang - The language code, e.g. en
Valid for Enterprise License holders with multilingual bots, this action lets you change the bot's language for the current user. Digital Agent comes pre-packed with a translation engine that helps developers design a bot in one language while catering to users of all supported and configured languages.
Content Types
Content types are the primary way to display content in the chat.
Audio
The audio component can stream MP3 audio (.mp3). You could use it to stream music from Spotify or YouTube. This component could also stream audio files stored on your server.
Figure: Audio content
Card
A card is a single element of a carousel. It can be view as one card in a deck of cards. The carousel is the whole deck of card.
Figure: Card
Carousel
A carousel is a series of cards. This component is useful in e-commerce chatbot. The image element can be a URL or a Data URLs.
Figure: Carousel
Dropdown
A dropdown displays a list of choices.
Figure: Dropdown
Image
You can display images. You can use a URL image.
Figure: Image
Location
It displays a place on a map.
File
Digital Agent can display PDF file.
Figure: File
Single Choice
This component carries a message, usually a question, and suggests choices to the user to fulfill the message. The user can only pick one option, and on selecting the preference, you can instruct your chatbot to get a custom value.
Figure: Single Choice
Text
The text content type is a regular text message with optional typing indicators and alternates. You can use markdown in your text to add formatting and style, but please ensure that the target channel can render this text.
You can write HTML in the text content on the web channel, and your chatbot will render it correctly. This opens up the possibility of including iFrames and constructing miniature web pages (commonly known as web views) in your content without creating custom components.
Figure: Text
Video
You can either upload a video or link to a video file that will be fetched when the content element is invoked.
Figure: Video