Embabel Agent Documentation/com.embabel.agent.api.validation.guardrails/UserInputGuardRail

UserInputGuardRail

interface UserInputGuardRail : GuardRail

Validates user inputs before LLM execution.

UserInputGuardRails provide safety checks on content that users provide to AI systems, ensuring that potentially harmful, inappropriate, or policy-violating content is detected and handled before being processed by the LLM.

This interface provides overloads for different input types while maintaining the base string validation from the GuardRail interface.

Properties

description

abstract val description: String

Description of what this validator checks. Used for documentation and debugging purposes.

name

abstract val name: String

Human-readable name for this validator. Used for logging, error reporting, and configuration.

Functions

combineMessages

open fun combineMessages(userMessages: List<UserMessage>): String

Combines multiple user messages into a single string for validation.

validate

abstract fun validate(input: String, blackboard: Blackboard): ValidationResult

Validate the given input within the provided blackboard context.

open fun validate(content: MultimodalContent, blackboard: Blackboard): ValidationResult

Validate multimodal content containing text and potentially images.

open fun validate(userMessages: List<UserMessage>, blackboard: Blackboard): ValidationResult

Validate a list of user messages from a conversation.