Struct LLMMiniLoopExecutor 

pub struct LLMMiniLoopExecutor { /* private fields */ }

Production MiniLoopExecutor backed by a real LLM provider.

Uses a fast/cheap model for lightweight decisions such as task complexity classification, compression decisions, retry classification, and routing.

Implementations

impl LLMMiniLoopExecutor

pub fn new(provider: Arc<dyn LLMProvider>, model: String) -> Self

Trait Implementations

impl MiniLoopExecutor for LLMMiniLoopExecutor

fn decide<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, _session: &'life1 Session, prompt: &'life2 str, context: &'life3 str, ) -> Pin<Box<dyn Future<Output = Result<MiniLoopDecision, AgentError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Execute a fast decision using a cheap model.

fn evaluate_task<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, _session: &'life1 Session, _tool_calls: &'life2 [ToolCall], _round: usize, ) -> Pin<Box<dyn Future<Output = Result<MiniLoopDecision, AgentError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Evaluate task progress using a fast model.