Struct VeilClient 

pub struct VeilClient { /* private fields */ }

Async client for the Mugen Veil verifiable inference gateway.

Construct via VeilClient::builder().

VeilClient is cheap to clone — the underlying reqwest::Client uses an Arc internally and shares the connection pool across clones.

Implementations

impl VeilClient

pub fn builder() -> VeilClientBuilder

Begin building a client. See VeilClientBuilder.

pub async fn health_check(&self) -> Result<Health>

GET /healthz — Returns gateway health information.

Use Health::is_healthy() to check overall readiness.

pub async fn submit_job( &self, model_id: impl Into<String>, input_data: Vec<Vec<f64>>, ) -> Result<String>

POST /v1/jobs — Submit an inference job and return immediately.

Returns the job_id. Use get_job to poll status, or verify_inference to submit and wait.

Arguments

• model_id — the model name registered with the gateway (e.g. "tiny_mlp_v1")
• input_data — row-major input tensor, e.g. vec![vec![0.1, 0.2, 0.3, 0.4]]

pub async fn get_job(&self, job_id: &str) -> Result<Job>

GET /v1/jobs/{id} — Poll the status of a job.

pub async fn get_proof(&self, job_id: &str) -> Result<Proof>

GET /v1/jobs/{id}/proof — Fetch the raw proof bytes for a completed job.

The gateway returns HTTP 202 if the job is not yet complete. Returns VeilError::Api with status 202 in that case — callers should poll get_job first.

pub async fn register_model( &self, req: RegisterModelRequest, ) -> Result<RegisterModelResponse>

POST /v1/models — Register an ONNX model artifact with the gateway.

Pins the artifact to IPFS and registers it on-chain. Requires PINATA_JWT to be configured on the gateway.

pub async fn verify_inference( &self, model_id: impl Into<String>, input_data: Vec<Vec<f64>>, ) -> Result<VerifyResult>

Submit an inference job and block until it reaches a terminal state.

Polls GET /v1/jobs/{id} at the configured poll_interval until the job status is one of done, settled, or failed, or until the configured timeout elapses.

Arguments

• model_id — registered model name, e.g. "tiny_mlp_v1"
• input_data — row-major input tensor

Errors

• VeilError::Timeout — polling exceeded the configured timeout
• VeilError::JobFailed — the gateway reported the job as failed
• VeilError::Api — gateway returned a non-2xx response
• VeilError::Http — network-level failure

Example

let client = VeilClient::builder()
    .base_url("http://localhost:8080")
    .build()?;

let result = client
    .verify_inference("tiny_mlp_v1", vec![vec![0.1, 0.2, 0.3, 0.4]])
    .await?;

println!("tx_hash: {:?}", result.tx_hash);

Trait Implementations

impl Clone for VeilClient

fn clone(&self) -> VeilClient

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for VeilClient

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

Formats the value using the given formatter.