Intent export and import

In new projects, you can reuse intents created in other projects by using the export and import features.

Export

To export all intents from the project, navigate to Intents and click above the list of intents. A JSON file will be downloaded, which can be used for importing intents to a new project.

tip

You can also export the entire project by clicking → Download next to the project name. In the downloaded archive, the nlu.json file contains NLU data, including intents and their settings.

Import

To import intents, navigate to the Intents section and click Import on the top panel. Select the necessary file and what should be done with existing intents and intents having duplicate names. Click Import again.

tip

The file contains an array of intents with all their configured parameters, slot filling settings, and training phrases.

File structure

Intents

Objects representing each intent contain the following properties:

Property	Type	Description	Required
id	Number	Intent ID	No
path	String	Intent path	Yes
description	String	Intent description	No
answer	String	Default answer	No
enabled	Boolean	True when the intent is enabled	No
phrases	Array of objects with a text property	Training phrases	Yes
patterns	Array of strings	Training patterns	No
slots	Array of objects	Slots	No

Slots

The slots property contains an array objects describing slots — entities that are requested during slot filling. Every object contains the following properties:

Property	Type	Description
name	String	Slot name
entity	String	The corresponding entity name
required	Boolean	True when the slot is required
prompts	Array of strings	Clarifying questions
array	Boolean	True when the slot is processed as an array

File example

The imported file must be in JSON format. The following example illustrates a valid file:

[
    {
        "id": 12345,
        "path": "/Callback",
        "enabled": true,
        "shared": false,
        "phrases": [
            {
                "text": "Could you call me later"
            },
            {
                "text": "Call me back @duckling.time"
            }
        ],
        "classificationRules": {
            "refs": []
        },
        "patterns": [],
        "slots": [
            {
                "name": "dateTime",
                "entity":"duckling.time",
                "required": true,
                "prompts": [
                    "When should I call you back?",
                    "When will it be best to call?"
                ],
                "array": false
            }
        ],
        "priority": 0
    }
]