Struct AgentOrchestrator

Source

pub struct AgentOrchestrator { /* private fields */ }

Expand description

Central orchestrator for agent lifecycle and workflows

The AgentOrchestrator manages the execution of agents, including:

Agent lifecycle (initialization, execution, cleanup)
Inter-agent communication and handoff
Sequential, parallel, and conditional workflows
Error handling and retry logic

§Examples

use ricecoder_agents::{AgentOrchestrator, AgentRegistry, AgentTask, TaskType, TaskTarget, TaskScope};
use std::sync::Arc;
use std::path::PathBuf;

#[tokio::main]
async fn main() {
    let registry = Arc::new(AgentRegistry::new());
    let orchestrator = AgentOrchestrator::new(registry);

    let task = AgentTask {
        id: "task-1".to_string(),
        task_type: TaskType::CodeReview,
        target: TaskTarget {
            files: vec![PathBuf::from("src/main.rs")],
            scope: TaskScope::File,
        },
        options: Default::default(),
    };

    let results = orchestrator.execute(vec![task]).await.unwrap();
}

Implementations§

Source §

impl AgentOrchestrator

Source

pub fn new(registry: Arc<AgentRegistry>) -> Self

Create a new agent orchestrator

§Arguments

registry - The agent registry containing registered agents

§Returns

A new AgentOrchestrator with default retry configuration

Source

pub fn with_retry_config( registry: Arc<AgentRegistry>, retry_config: RetryConfig, ) -> Self

Create a new agent orchestrator with custom retry configuration

§Arguments

registry - The agent registry containing registered agents
retry_config - Custom retry configuration

§Returns

A new AgentOrchestrator with the specified retry configuration

Source

pub fn set_retry_config(&mut self, retry_config: RetryConfig)

Set the retry configuration

Source

pub fn retry_config(&self) -> &RetryConfig

Get the retry configuration

Source

pub async fn execute_with_retry( &self, tasks: Vec<AgentTask>, ) -> Result<Vec<AgentOutput>>

Execute agents for the given tasks with retry logic

This method executes the given tasks with automatic retry on failure. If execution fails, it will retry up to max_retries times with exponential backoff.

§Arguments

tasks - Vector of tasks to execute

§Returns

A Result containing the agent outputs or an error

Source

pub async fn execute(&self, tasks: Vec<AgentTask>) -> Result<Vec<AgentOutput>>

Execute agents for the given tasks

This method executes the given tasks using the orchestrator’s scheduler to determine execution order and parallelism. Tasks are executed according to their dependencies, with independent tasks running in parallel.

§Arguments

tasks - Vector of tasks to execute

§Returns

A Result containing the agent outputs or an error

Source

pub async fn execute_and_aggregate( &self, tasks: Vec<AgentTask>, ) -> Result<AgentOutput>

Execute and aggregate results from multiple agents

This method executes the given tasks and then aggregates all results into a single AgentOutput, combining findings, suggestions, and generated content.

§Arguments

tasks - Vector of tasks to execute

§Returns

A Result containing the aggregated agent output or an error

Source

pub async fn execute_conditional<F>( &self, tasks: Vec<AgentTask>, condition: F, ) -> Result<Vec<AgentOutput>>
where F: Fn(&[AgentOutput]) -> bool,

Execute tasks conditionally based on a predicate

This method supports conditional workflows where later tasks only execute if earlier tasks meet certain conditions.

§Arguments

tasks - Vector of tasks to execute
condition - Closure that determines if execution should continue Takes the current outputs and returns true to continue, false to stop

Source