Card properties

The following properties is possible to add on each card regardless of the type

Ask again

The ask_again property is only valid for input cards and lets you define if you would like to always ask for the inputted value every time the scenario is triggered or if you like to have it stored in memory. As default Ebbot will never ask for a variable that it already has.

[
	{
		"name": "state_1613396374509",
		"component": "ebbot_input_string",
		"properties": {
			"text": "Tell me your name please",
			"output": "name",
			"ask_again": true
		}
	}
]

Database

The database object is used to add data to the database.

"database":{ 
    "first_name": "Emmett",
    "last_name": "Brown"
}

You can define whatever property you like and it's then reachable through either {{ property }} or data['db']['property']

Reaction

With the reaction property you can assign a card as a reaction to an output. A card that has a reaction will only be played the first time the variable is set. This is useful if you would like to add a thank you response when collecting values for instance.

Here is an example of a reaction that says "Thank you" once the user has entered their name. The second time the scenario is triggered none of the cards will be played.

[
	{
		"name": "state_1613396374509",
		"component": "ebbot_input_string",
		"properties": {
			"text": "Tell me your name please",
			"output": "name",
			"ask_again": false,
			"validation": {}
		}
	},
	{
		"name": "state_1613396382298",
		"component": "ebbot_text",
		"reaction": "name",
		"properties": {
			"text": "Thank you"
		}
	}
]

Stop

By using the stop property you can discard all the remaining cards in a scenario. By default this property is set to false.

[
	{
		"name": "state_1613396374509",
		"component": "ebbot_text",
		"properties": {
			"text": "Hello there 😊",
		},
		"stop": true
	}
]

Next

By using the next property you can link to another scenario at a given state. It is useful if you would like to trigger another scenario without using buttons. You can make a trigger to a scenario or to a given state in that scenario.

[
	{
		"name": "state_1613396374509",
		"component": "ebbot_text",
		"properties": {
			"text": "Hello there 😊",
		},
		"next": {
		"scenario": "scenario_name_or_id",
		"state": "state_id" ## OPTIONAL
	}
]

Message visibility

The Message Visibility toggle controls whether the content of a specific card is included in the data sent to EbbotGPT. How it Works

• Enabled (On): The message is sent to the LLM and becomes part of the conversation history used to generate a response.

• Disabled (Off): The LLM never "sees" this card. It remains part of the flow for the end-user (if applicable) but is completely ignored by the AI's logic.

    "visibleTo": [
      "user",
      "agent"
    ]

GPT message role

The Message Role setting allows you to define how EbbotGPT (the LLM) perceives a specific content card within a scenario. By manually selecting a role, you can "force" the LLM to interpret a message as coming from a specific source. Available Roles

• User: The LLM treats the message as an input or question from the end-user.

• Assistant: The LLM treats the message as its own previous response.

• System: The LLM treats the message as a high-level instruction or background rule it must follow.

Why use this?

This feature gives you granular control over the conversation history. For example:

• Simulate History: You can add a text card and set it to User. Even though the AI agent technically sent the text within the flow, EbbotGPT will believe the user was the one who said it and respond accordingly.

• Inject Context: Use the System role to indicate that the message was sent by a static scenario or workflow and not by the AI itself.

The LLM relies on these roles to understand the "who’s who" of a conversation. Changing these could significantly impact how the AI generates its next response.

    "gptRole": "assistant"