Struct RetryProvider 

pub struct RetryProvider {
    pub max_retries: u32,
    pub base_delay: Duration,
    /* private fields */
}

Opt-in retry wrapper for any LlmProvider.

Wraps an inner provider and retries transient errors (timeout, network, HTTP 500/429) up to max_retries times with exponential backoff starting from base_delay. Non-retryable errors (auth, process, nested session, other HTTP status codes) are returned immediately.

Implements LlmProvider itself, making it transparent to consumers.

Fields

max_retries: u32

Maximum number of retry attempts after the first failure.

base_delay: Duration

Delay between retry attempts.

Implementations

impl RetryProvider

pub fn new(inner: Arc<dyn LlmProvider>) -> Self

Creates a new RetryProvider with default settings (3 retries, 1s delay).

Parameters

• inner: The provider to wrap with retry logic.

pub fn with_config( inner: Arc<dyn LlmProvider>, max_retries: u32, base_delay: Duration, ) -> Self

Creates a new RetryProvider with custom retry settings.

Parameters

• inner: The provider to wrap with retry logic.
• max_retries: Maximum retry attempts after the initial failure.
• base_delay: Initial delay between retries; doubles on each subsequent attempt.

Trait Implementations

impl LlmProvider for RetryProvider

fn complete<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, system_prompt: &'life1 str, user_prompt: &'life2 str, config: &'life3 CompletionConfig, ) -> Pin<Box<dyn Future<Output = Result<String, ProviderError>> + Send + 'async_trait>>

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

Sends a completion request to the LLM provider.

fn name(&self) -> &str

Returns the provider’s name (e.g., “claude”, “claude-cli”, “openai”).

fn model(&self) -> &str

Returns the model identifier (e.g., “claude-sonnet-4-6”).