Struct SelfHealingEngine 

pub struct SelfHealingEngine {
    pub config: HealingConfig,
    /* private fields */
}

The self-healing engine tracks failures, manages circuit breakers, and provides retry-with-backoff for agent tasks.

This is the central coordination point for all self-healing in RavenClaws. It is shared across the agent loop, swarm orchestrator, heartbeat agent, and background task manager via Arc<RwLock<>>.

Fields

config: HealingConfig

Configuration

Implementations

impl SelfHealingEngine

pub fn new() -> Self

Create a new self-healing engine with default configuration.

pub fn with_config(config: HealingConfig) -> Self

Create a new self-healing engine with custom configuration.

pub fn record_failure(&mut self, agent_id: &str, error: &str) -> u32

Record a failure for an agent.

Returns the updated failure count for that agent.

pub fn record_success(&mut self, agent_id: &str)

Record a success for an agent — resets failure tracking.

pub fn is_healthy(&mut self, agent_id: &str) -> bool

Check if an agent is healthy.

An agent is healthy if:

1. Its circuit breaker allows execution (Closed or HalfOpen)
2. It’s not marked as dead

pub fn failure_count(&self, agent_id: &str) -> u32

Get the failure count for an agent.

pub fn circuit_state(&self, agent_id: &str) -> Option<HealingCircuitState>

Get the circuit state for an agent.

pub fn dead_workers(&self) -> Vec<String>

Get a list of agent IDs that are marked as dead and past the replacement timeout.

pub fn all_dead_workers(&self) -> Vec<String>

Get a list of all agent IDs that are currently dead (regardless of timeout).

pub fn delay_for_attempt(&self, attempt: u32) -> Duration

Calculate delay for a retry attempt using exponential backoff with jitter.

pub async fn retry_with_backoff<F, T>( &self, agent_id: &str, is_transient: impl Fn(&str) -> bool, operation: F, ) -> Result<T, String>

where F: FnMut(u32) -> Result<T, String>,

Execute an async function with retry and exponential backoff.

The function receives the attempt number (0-based) and should return Ok(T) on success or Err(String) on failure. Only transient errors should be retried — pass is_transient to classify.

Returns Ok(T) on success, or Err(String) with the last error after all retries are exhausted.

pub fn reset(&mut self)

Reset all failure records and circuit breakers.

pub fn reset_agent(&mut self, agent_id: &str)

Reset failure records for a specific agent.

pub fn tracked_agents(&self) -> usize

Get the total number of tracked agents.

pub fn dead_agent_count(&self) -> usize

Get the number of dead agents.

pub fn open_circuit_count(&self) -> usize

Get the number of agents with open circuit breakers.

Trait Implementations

impl Debug for SelfHealingEngine

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for SelfHealingEngine

fn default() -> Self

Returns the “default value” for a type.