Struct ProviderConfig 

pub struct ProviderConfig {
    pub name: String,
    pub model: String,
    pub base_url: Option<String>,
    pub api_key: Option<String>,
    pub retry: Option<RetryProviderConfig>,
    pub prompt_caching: bool,
    pub cascade: Option<CascadeConfig>,
    pub circuit: ProviderCircuitConfig,
}

LLM provider configuration.

When running as a cloud-delegated runtime (daemon mode with no agents), the provider section can be omitted — per-request provider keys are used instead.

Fields

name: String

model: String

base_url: Option<String>

Custom API endpoint URL (overrides the default for the provider). Useful for self-hosted models, Azure, or proxies.

api_key: Option<String>

Direct API key (alternative to environment variable). Prefer env vars in production; this is for testing/local dev.

retry: Option<RetryProviderConfig>

Retry configuration for transient LLM API failures.

prompt_caching: bool

Enable Anthropic prompt caching (system prompt + tool definitions). Only effective for the anthropic provider. Defaults to false.

cascade: Option<CascadeConfig>

Model cascading configuration. When enabled, tries cheaper models first and escalates to the main model only when the confidence gate rejects.

circuit: ProviderCircuitConfig

Circuit breaker configuration for this provider. When absent, sensible defaults are used (5 failures → 30 s open, max 300 s).

Trait Implementations

impl Debug for ProviderConfig

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

Formats the value using the given formatter.

impl Default for ProviderConfig

fn default() -> ProviderConfig

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for ProviderConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.