Struct LlmConfig 

pub struct LlmConfig {

    pub url: String,
    pub model: String,
    pub api_key: Option<String>,
    pub headers: HashMap<String, String>,
    pub timeout: Duration,
    pub temperature: f64,
    pub thinking: Option<bool>,
    pub model_params: HashMap<String, Value>,
    pub cache: bool,
    pub max_attempts: u32,
    pub provider: Provider,
    pub deployment: Option<String>,
    pub api_version: Option<String>,
    pub auth: AuthConfig,
    pub header_commands: HashMap<String, String>,
    pub aws: AwsConfig,
}

Configuration for the LLM client — bundles URL, model, auth, timeouts, and provider-specific options into a single struct passed everywhere.

Fields

url: String

OpenAI-compatible API base URL (without trailing /v1/…).

model: String

Model name (e.g. gpt-4o-mini, deepseek).

api_key: Option<String>

API key sent as Authorization: Bearer <key>.

headers: HashMap<String, String>

Custom headers appended to every LLM request.

timeout: Duration

HTTP timeout.

temperature: f64

Sampling temperature (0.0–1.0).

thinking: Option<bool>

Enable extended thinking / reasoning tokens. None = don’t send any thinking key (provider default).

model_params: HashMap<String, Value>

Provider-specific parameters merged into the request body (e.g. effort = "high" for Anthropic).

cache: bool

Send provider-side prompt-cache markers (default true). Applies to providers that need explicit markers: AWS Bedrock (cachePoint) and Anthropic-style OpenAI-compatible models (cache_control). Set to false to disable.

max_attempts: u32

How many times a single call to this endpoint is retried on transient failures before giving up (or moving to the next fallback endpoint). Default 3; override globally with HARNESS_LLM_CALL_ATTEMPTS.

provider: Provider

LLM provider protocol (defaults to OpenAI-compatible).

azure switches to the Azure OpenAI deployments endpoint, bedrock to the SigV4-signed AWS Bedrock Converse API (feature aws).

deployment: Option<String>

Azure OpenAI deployment name (Provider::Azure). Defaults to model when unset.

api_version: Option<String>

Azure OpenAI API version (Provider::Azure). Defaults to 2024-10-21.

auth: AuthConfig

Authentication configuration (API key, token command, Entra ID).

header_commands: HashMap<String, String>

Extra HTTP headers produced by running a command per call, keyed by header name. Provider-agnostic.

aws: AwsConfig

AWS credential settings (Provider::Bedrock).

Implementations

impl LlmConfig

pub fn from_env() -> Self

Build a config from environment defaults, falling back to safe values when no env vars are set.

Trait Implementations

impl Clone for LlmConfig

fn clone(&self) -> Self

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for LlmConfig

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

Formats the value using the given formatter.

impl Default for LlmConfig

fn default() -> Self

Returns the “default value” for a type.