Struct AgentLoop 

pub struct AgentLoop { /* private fields */ }

The agent loop is the core processing engine

Implementations

impl AgentLoop

pub fn register_default_tools(&mut self, tool_config: ToolConfig)

Register default tools (for use after construction)

impl AgentLoop

pub fn new( bus: MessageBus, provider: Arc<dyn LLMProvider>, workspace: PathBuf, model: Option<String>, max_iterations: Option<usize>, ) -> Self

Create a new agent loop

pub fn with_tools( bus: MessageBus, provider: Arc<dyn LLMProvider>, workspace: PathBuf, model: Option<String>, max_iterations: Option<usize>, tool_config: ToolConfig, runtime_control_rx: Option<UnboundedReceiver<RuntimeControlCommand>>, ) -> Self

Create a new agent loop with tool configuration

pub async fn run(&mut self) -> Result<(), Box<dyn Error>>

Run the agent loop, processing messages from the bus

pub async fn process_inbound_message( &mut self, msg: InboundMessage, event_tx: Option<&UnboundedSender<AgentEvent>>, ) -> Result<Option<OutboundMessage>, Box<dyn Error>>

Process a single inbound message

pub async fn process_direct( &mut self, content: impl Into<String>, _session_key: impl Into<String>, channel: impl Into<String>, chat_id: impl Into<String>, ) -> Result<String, Box<dyn Error>>

Process a message directly (for CLI or testing)

pub async fn process_direct_stream( &mut self, content: impl Into<String>, _session_key: impl Into<String>, channel: impl Into<String>, chat_id: impl Into<String>, event_tx: UnboundedSender<AgentEvent>, ) -> Result<String, Box<dyn Error>>

Process a message directly and emit streaming events for UI consumers.