Automation JSON structure

When you export an axiom.ai automation, you get a JSON file containing the inner workings of the automation. The export is useful for sharing automations with the support team, with a colleague, or with anyone you're collaborating with. Most of the file is data axiom.ai uses to reconstruct the automation, but some of it is interesting on its own. This page documents the most useful fields. We've left out fields that relate to deprecated features, internal extension state, or rarely-used settings.

You can think of an exported automation as a snapshot of the automation's current state, including every step variable, error setting, and configuration value.

Warning: Don't edit an exported automation by hand. Errors in the JSON prevent the file from being imported back into the extension.

Identifiers

The first few fields in the JSON identify the automation:

"id": 938717,
"name": "Send to Supabase",
"templateId": 0,
"triggers": [ ... ]

If the automation has a schedule, the triggers array contains an entry like this:

{
    "id": 10126,
    "interval_type": "seconds",
    "name": "",
    "starting_time": "2025-02-26 16:26:23+00",
    "status": "ready",
    "time_criteria": 86400,
    "type": "schedule"
}

Data

The data key holds everything about the automation's behaviour: steps, settings, and the configurations that drive each run. The top of the data object looks like this:

"builderTemplateId": 0,
"context": [{ "context": "url", "url": "" }],
"continueOnError": true,
"disablePageChangeMonitoring": true,
"executablePath": "...",
"extensionToLoad": "...",
"form": [ ... ]

Step data

Step data lives in the form key. It's an array of step objects shaped like this:

{
    "description": "Instruct the bot to go to a new page.",
    "hasErrors": false,
    "index": 0,
    "machine_name": "WidgetDriverGoto",
    "metadata": "",
    "name": "google.com",
    "original_name": "Go to page",
    "params": [ ... ],
    "stepNumber": "1",
    "token": ""
}

Note: The description and parameter description fields are stored at creation time so older steps continue to make sense if the canonical step description changes in a later release.

Params

The params array inside each step object holds every parameter the step has, plus the value you've configured. For example, a single parameter on the Read data from a Google Sheet step looks like this:

{
    "collapsible": 0,
    "configurable": true,
    "default_value": "",
    "description": ["Select a google sheet to read from, or paste its URL here."],
    "help": [],
    "image": "",
    "name": "Spreadsheet",
    "sheetName": "Your Google Sheet",
    "type": "spreadsheet_picker",
    "value": "..."
}

Each parameter type may include extra fields specific to that parameter or step. The example above is a single parameter; most steps have several.

Automation settings

The remaining keys are the automation's settings, the same ones you configure from the Cog icon in the Builder.

"headless": false,
"iframesAllowed": true,
"incognitoMode": true,
"injector": { "injector": "dom", "magic_btn_status": true, "selector": "" },
"maximumRuntime": 600,
"mode": "browser",
"notifications": { ... },
"proxyAuth": null,
"storeCookies": true,
"templateId": 0,
"templateItem": null,
"templateMode": false,
"templateStarted": false,
"tokens": null,
"useLocalTimezone": true,
"userDataDir": "...",
"viewport": null

Notifications

The notifications key holds the email and webhook notification configuration:

{
    "emailAddress": "...",
    "emailOn": true,
    "onFailure": true,
    "onSuccess": false,
    "webhookOn": false,
    "webhookPayload": "",
    "webhookURL": ""
}

Wrapping up

Understanding the JSON structure of an automation can help when you need to share an automation with support or read someone else's exported file. We don't recommend editing an automation through the JSON, but if you do, work from the table above to know what each key controls. Any malformed JSON will prevent the file from being imported back into the extension.