Configure exit points and custom functions

← Back to table of contents

Exit points are defined locations in the Contoso Platform’s processing pipeline where you can insert custom logic without modifying the core product code. This topic describes how to create and register a custom function at an exit point.

Available exit points

Exit point Called when Typical use
BeforeIngestion Immediately after the platform reads a record from the source feed, before it applies field mapping. Normalize or clean up raw values.
AfterIngestion After field mapping, before detection rules evaluate the record. Enrich a record with a derived value, such as a customer segment.
BeforeScoring After detection rules run, before the platform combines scores. Adjust an individual score factor with custom logic.
AfterCaseCreation After the platform creates a case. Assign the case to a specific queue or team.

Create a custom function

  1. Open the Custom Functions page in the Contoso Admin Console.
  2. Select Add Function and choose the exit point to attach it to.

  3. Write the function body using the platform’s expression language. The following example runs at the AfterIngestion exit point and derives a simple risk flag from the transaction amount:

    function EnrichTransaction(record):
    if record.transaction.amount > 10000:
        record.customFields.highValueFlag = true
    else:
        record.customFields.highValueFlag = false
    return record
  4. Save and activate the function.
  5. Test the function with a sample record before you enable it in a production environment. See Test and validate your configuration.

Best practices

  • Keep custom functions small and focused on a single task.
  • Reference a configurable parameter instead of embedding business thresholds directly in custom function code. See Configure business logic.
  • Document the purpose of each custom function in its description field so future implementers understand why it exists.

Note: The function name and logic above are illustrative examples only and don’t reflect any real detection methodology.

← Previous: Set up data integration Next: Configure business logic →