Injection

Detect prompt injection and instruction-override style attacks.

YAML key: injection
Direction: input

Detect prompt injection and instruction-override style attacks.

Configuration

Prop

Type

defend.config.yaml (fragment)

guards:
  input:
    modules:
      - injection:
          strict: true

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

Modules

Guard modules add provider prompt fragments for input or output; configure them under guards.input.modules and guards.output.modules.

Jailbreak

Detect jailbreak and safety-bypass attempts.