Guardrails

Guardrails are scanners used to ensure safety, stability, and performance. Input and Output scanners in an Agentic App help monitor and regulate the data that the agents receive and generate.

For instance, to prevent the agent from processing inputs that contain sensitive or harmful content, you can add a regex input scanner. This scanner validates user inputs against predefined regular expression (regex) patterns, ensuring that only acceptable content is processed.

You can add scanners for both input and output. Learn more about Guardrails and how to add scanners.