❌ Error Handling

AgentTech SDKs use typed errors and sentinel values for precise error matching.

Error Types

Returned for HTTP 4xx/5xx responses from the API.

if (err instanceof PayApiError) {
  console.log(`HTTP ${err.statusCode}: ${err.body}`);
}

var reqErr *pay.RequestError
if errors.As(err, &reqErr) {
    log.Printf("HTTP %d: %s", reqErr.StatusCode, reqErr.Body)
}

Returned when the SDK rejects a request before it reaches the API (e.g. empty intent ID). Wraps a sentinel error for errors.Is matching.

var valErr *pay.ValidationError
if errors.As(err, &valErr) {
    log.Printf("Invalid input: %s", valErr.Message)
}

Wraps unexpected internal errors (JSON marshal failure, request creation, etc.).

var unexpErr *pay.UnexpectedError
if errors.As(err, &unexpErr) {
    log.Printf("Unexpected: %v", unexpErr.Err)
}

Use errors.Is (Go) or error type checking to match specific validation failures.

Sentinel	Meaning
`ErrEmptyBaseURL`	baseURL was empty in NewClient
`ErrEmptyIntentID`	intentID was empty
`ErrEmptySettleProof`	settleProof was empty in SubmitProof
`ErrMissingAuth`	ExecuteIntent called without auth
`ErrNilParams`	params argument was nil

if errors.Is(err, pay.ErrEmptyIntentID) {
    // intent ID was empty
}

Status Code	Meaning
`400`	Bad request — invalid parameters, amount out of range, or malformed input
`401`	Unauthorized — missing or invalid credentials
`403`	Forbidden — insufficient permissions for this operation
`404`	Not found — intent does not exist
`429`	Rate limited — too many requests (60 req/min/IP typical)
`503`	Service unavailable — temporary backend issue

The API allows approximately 60 requests per IP per minute. On HTTP 429, implement exponential backoff:

var reqErr *pay.RequestError
if errors.As(err, &reqErr) && reqErr.StatusCode == 429 {
    time.Sleep(backoff)
    // retry
}