Struct Synth 

pub struct Synth { /* private fields */ }

Main Synth AI client.

This is the primary entry point for interacting with Synth AI APIs.

Example

use synth_ai::Synth;

let synth = Synth::from_env()?;

// Or with explicit credentials
let synth = Synth::new("sk_live_...", None)?;

Implementations

impl Synth

pub fn new(api_key: impl Into<String>, base_url: Option<&str>) -> Result<Self>

Create a new Synth client with explicit credentials.

Arguments

• api_key - Your Synth API key
• base_url - Optional custom API base URL

pub fn from_env() -> Result<Self>

Create a client from the SYNTH_API_KEY environment variable.

Examples found in repository

examples/container_list.rs (line 5)

async fn main() -> Result<(), synth_ai::Error> {
    let synth = Synth::from_env()?;
    let deployments = synth.list_container_deployments().await?;
    println!("deployments={}", deployments.len());
    for dep in deployments {
        println!(
            "{} {} {} {}",
            dep.deployment_id, dep.name, dep.status, dep.container_url
        );
    }
    Ok(())
}

pub fn api_key_masked(&self) -> String

Get the API key (masked for display).

pub fn base_url(&self) -> &str

Get the base URL.

pub fn core(&self) -> &SynthClient

Access the underlying core client.

pub fn environment_pools(&self) -> Result<EnvironmentPoolsClient>

Create an Environment Pools client.

pub fn optimize(&self) -> OptimizeBuilder

Start a prompt optimization job.

Returns a builder to configure the optimization.

pub fn eval(&self) -> EvalBuilder

Start an evaluation job.

Returns a builder to configure the evaluation.

pub async fn tunnel( &self, port: u16, backend: TunnelBackend, ) -> Result<TunnelHandle>

Open a tunnel to a local port.

Arguments

• port - Local port to tunnel
• backend - Tunnel backend (cloudflare_managed_lease recommended)

pub async fn submit_gepa(&self, request: GepaJobRequest) -> Result<String>

Submit a raw GEPA job.

pub async fn submit_mipro(&self, request: MiproJobRequest) -> Result<String>

Submit a raw MIPRO job.

pub async fn get_job_status(&self, job_id: &str) -> Result<PromptLearningResult>

Get job status.

pub async fn poll_job( &self, job_id: &str, timeout_secs: f64, interval_secs: f64, ) -> Result<PromptLearningResult>

Poll job until complete.

pub async fn cancel_job(&self, job_id: &str, reason: Option<&str>) -> Result<()>

Cancel a job.

pub async fn pause_job(&self, job_id: &str, reason: Option<&str>) -> Result<()>

Pause a job.

pub async fn resume_job(&self, job_id: &str, reason: Option<&str>) -> Result<()>

Resume a paused job.

pub async fn complete( &self, request: GraphCompletionRequest, ) -> Result<GraphCompletionResponse>

Run graph completion.

pub async fn list_graphs( &self, kind: Option<&str>, limit: Option<i32>, ) -> Result<Value>

List registered graphs.

pub async fn verify( &self, trace: Value, rubric: Value, options: Option<VerifierOptions>, ) -> Result<VerifierResponse>

Run verifier on a trace.

pub async fn rlm_inference( &self, query: &str, context: Value, options: Option<RlmOptions>, ) -> Result<Value>

Run RLM (Retrieval-augmented LM) inference.

pub fn graph_evolve(&self) -> GraphEvolveClient<'_>

Create a Graph Evolve client for advanced operations.

pub fn graph_evolve_job_from_payload( &self, payload: Value, ) -> Result<GraphEvolveJob>

Create a Graph Evolve job from a payload.

pub fn graph_evolve_job_from_id(&self, job_id: &str) -> Result<GraphEvolveJob>

Reconnect to a Graph Evolve job by ID.

pub async fn verify_rubric( &self, trace: Value, rubric: Value, ) -> Result<VerifierResponse>

Verify a trace against a rubric with default options.

pub fn container_client( &self, base_url: &str, api_key: Option<&str>, ) -> ContainerClient

Create a Container container client.

pub async fn deploy_container_from_dir( &self, spec: ContainerDeploySpec, context_dir: impl AsRef<Path>, wait_for_ready: bool, build_timeout_s: f64, ) -> Result<ContainerDeployResponse>

Deploy a managed Container from a context directory.

pub async fn list_container_deployments( &self, ) -> Result<Vec<ContainerDeploymentInfo>>

List managed Container deployments for the current org.

Examples found in repository

examples/container_list.rs (line 6)

async fn main() -> Result<(), synth_ai::Error> {
    let synth = Synth::from_env()?;
    let deployments = synth.list_container_deployments().await?;
    println!("deployments={}", deployments.len());
    for dep in deployments {
        println!(
            "{} {} {} {}",
            dep.deployment_id, dep.name, dep.status, dep.container_url
        );
    }
    Ok(())
}

pub async fn get_container_deployment( &self, deployment_id: &str, ) -> Result<ContainerDeploymentInfo>

Fetch a managed Container deployment by ID.

pub async fn get_container_deployment_status( &self, deployment_id: &str, ) -> Result<ContainerDeployStatus>

Fetch managed Container deployment status by ID.

pub async fn eval_results(&self, job_id: &str) -> Result<Value>

Fetch detailed eval results.

pub async fn download_eval_traces(&self, job_id: &str) -> Result<Vec<u8>>

Download eval traces as ZIP bytes.

pub fn trace_uploader(&self) -> Result<TraceUploadClient>

Create a trace uploader for large traces.

pub async fn upload_trace( &self, trace: Value, expires_in_seconds: Option<i64>, ) -> Result<String>

Upload a trace and return its trace_ref.

pub async fn stream_job_with_callback<F>( &self, job_id: &str, endpoints: StreamEndpoints, callback: F, ) -> Result<Value>

where F: Fn(&StreamMessage) + Send + Sync + 'static,

Stream job events with a callback and return final status.

Trait Implementations

impl Debug for Synth

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

Formats the value using the given formatter.