Skip to content

Enable Input/Output Scanners

Input and output scanners help evaluate prompts and responses within your application.

  • Input scanners assess user inputs or prompts sent to an AI node.
  • Output scanners evaluate the responses returned from the LLM.

Note

All system scanners are pre-deployed and available by default. To use them, you must enable the required scanners in your agentic app or workflow tool.

Steps to enable scanners:

  1. Log in to your account and open the Guardrails settings in your Agentic app or Tool:

    • For Agentic apps: Go to Agentic apps, select your app, then go to Settings > PII & Guardrails, and select the Guardrails tab.
    • For Tools: Go to Tools, select the tool you want to configure, and then select Guardrails.

  2. On the Guardrails page, review the Input scanners and Output scanners tabs.
    Turn on the toggle next to each pre-deployed scanner you want to apply. Enable Guardrail

  3. To configure scanner settings, click the scanner you enabled, adjust the available options, and then click Save.
    The available options vary by scanner type. For example, the Toxicity scanner includes options such as Risk Threshold and Detection Sensitivity, while the Regex scanner includes options such as Scanner mode (Block or Allow), pattern entry fields for defining regex rules, and a Risk threshold slider. Configure Settings