You can transform and export data directly from Cerb into common formats expected by other apps and services (e.g. JSONL for training Hugging Face models).

In this example we’ll export information about PNG image attachments on the past 90 days of tickets.

From Search » Attachments, run a search query to filter an attachment worklist.

For instance, PNG images attached to messages within tickets created in the past 90 days:

name:*.png on.message:(ticket:(created:"-90 days"))

Click on the export icon in the top right of the worklist.

Select an output format in Export rows as:. Common formats are CSV or JSONL.

Select the Build tab.

You can use Export KATA to define a custom export schema, allowing you to format or generate fields using scripting.

For instance, the following Export KATA includes:

• A creation date (as a Unix timestamp)
• The URL of the first ticket linked to each attachment
• The filename

column/updated:
  label: Created

column/ticket_url:
  label: Ticket URL
  value@raw: {{on|filter((record) =>'message' == record._type)|first.ticket_record_url}}

column/name:
  label: Filename

The value@raw: annotation is used to delay evaluation of placeholders until each row is exported. Without the @raw annotation, a placeholder would be evaluated when the KATA is first read and its value would be the same for each record.

Click on the Export button to export the worklist results to your desired format using the schema.

You can optionally use the |date filter to convert the date into a human-readable format:

column/updated:
  label: Created
  value@raw: {{updated|date('Y-m-d H:i:s')}}

You can find the dictionary placeholders for each record type in the docs.