Enum RigAgent 

pub enum RigAgent {
    OpenAi {
        agent: Agent<RetryingModel<CompletionModel>>,
        tool_call_max: usize,
    },
}

Runtime-dispatched Rig agent. The OpenAi-backed and mistralrs-backed CompletionModel impls produce concretely different Agent<M> types (Rig’s trait carries associated types, so a single concrete RigAgent can’t carry both). Callers (the agent loop) match on the variant.

Variants

OpenAi

Fields

agent: Agent<RetryingModel<CompletionModel>>

tool_call_max: usize

Implementations

impl RigAgent

pub async fn run_turn( &self, prompt: &str, history: &mut Vec<Message>, ) -> Result<String>

Run one user turn: prompt the model, drive the model->tool->model loop, extend history with everything emitted (user prompt + tool turns + final assistant reply), and return the assistant’s text reply.

The per-turn OutrigPromptHook prints [outrig] tool call: ... to stderr for every tool invocation and terminates the loop after the resolved tool-call max. If the hook terminates the loop, Rig returns the partial chat history it had accumulated; outrig splices in that new suffix so the user can send a follow-up prompt to continue.