Behaviour for the LLM that generates adversarial prompts.

Plugins call generate/3 with a meta-prompt and a NimbleOptions-style schema describing the expected attack list. Different implementations swap the underlying client without touching plugin code.

Configuration

config :tribunal, :red_team_attacker, Tribunal.RedTeam.Attacker.ReqLLM
config :tribunal, :red_team_attacker_model, "anthropic:claude-sonnet-4-5"

Default is Tribunal.RedTeam.Attacker.ReqLLM (used when req_llm is loaded). Tests typically use Tribunal.RedTeam.Attacker.Stub with seeded responses.

Implementing a custom attacker

defmodule MyApp.RedTeam.Attacker do
  @behaviour Tribunal.RedTeam.Attacker

  @impl true
  def generate(prompt, schema, opts) do
    # call your model, return {:ok, %{key => value}} or {:error, reason}
  end
end