Topic

Restrict user input to allowed topics.

YAML key: topic
Direction: input

Restrict user input to allowed topics.

Configuration

Prop

Type

defend.config.yaml (fragment)

guards:
  input:
    modules:
      - topic:
          allowed_topics: []

Configure under guards.input.modules when using an LLM-backed input provider.

Bias output

Detect biased or unfair statements in model output.

Topic output

Require model output to stay within allowed topics.