Struct ToolCoordinator 

pub struct ToolCoordinator { /* private fields */ }

Runtime that drives the LLM + tool-calling loop.

Wraps a backend and a tool registry, sends prompts with tool definitions, executes requested tools, feeds results back, and repeats until the model produces a final text response or a halt condition fires.

Example

use pawan::coordinator::{ToolCoordinator, ToolCallingConfig};
use pawan::tools::ToolRegistry;
use std::sync::Arc;

let backend = Arc::new(my_backend);
let registry = Arc::new(ToolRegistry::new());
let coordinator = ToolCoordinator::new(backend, registry, ToolCallingConfig::default());

let result = coordinator.execute(Some("You are helpful."), "What is 2+2?").await?;
println!("{}", result.content);

Implementations

impl ToolCoordinator

pub fn new( backend: Arc<dyn LlmBackend>, registry: Arc<ToolRegistry>, config: ToolCallingConfig, ) -> Self

Create a new ToolCoordinator.

pub async fn execute( &self, system_prompt: Option<&str>, user_prompt: &str, ) -> Result<CoordinatorResult>

Execute a tool-calling session starting from a plain prompt.

Builds an initial [system?, user] message list and drives the loop.

pub async fn execute_with_history( &self, messages: Vec<ConversationMessage>, ) -> Result<CoordinatorResult>

Execute a tool-calling session from an existing message history.

This is the primary loop: it calls the backend, dispatches tool calls, appends results to history, and repeats until the model emits a final text response or a halt condition fires.