Skip to main content
Use the HTTP status and structured error code to choose the next action. Correct authentication, permission, and validation problems before retrying. For rate limits and retryable service failures, respect the response’s retry guidance.

​
Rate-limit headers

Public API limits apply per key and per workspace: 120 requests per minute for an individual key and 600 per minute across a workspace’s keys. The versioned API rate-limit buckets apply these limits to their respective API version. Responses expose X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset. The reset value is a Unix timestamp in seconds. A 429 response includes Retry-After in seconds. Creating extra keys does not remove the workspace ceiling.

​
Error formats

Visibility errors normally contain error.code, error.message, optional error.details, and a top-level requestId. Data domain errors use error.code, error.requestId, error.retryable, error.retryAfterSeconds, error.fieldPath, error.parameters, and error.repairActions. Authentication and outer rate-limit failures on Data routes can use the visibility error envelope. Handle both shapes. Include the request reference when contacting support. Do not log the credential. Read safe writes before automatically retrying a request that creates or changes data.