Errors
Error reference
Public error codes, recovery decisions, and docs URLs for Everyn clients.
Everyn errors are meant to be recoverable and linkable. Public API errors include a stable code, a readable message, a requestId, a docsUrl, and optional fieldErrors. Dataset upload failures also expose public-safe failure metadata on the resource itself.
Use this section as the recovery hub. Use API errors when you need the exact HTTP envelope shape. Use specific error-code pages when an error includes a docsUrl.
Reading and recovering from errors
How to read an Everyn error
| Field | Meaning | What to do |
|---|---|---|
status | HTTP status category. | Decide whether this is validation, auth, state, retry, or internal. |
code | Stable machine-readable error. | Open the matching docs page when available. |
message | Readable summary. | Quote or paraphrase without losing the code. |
requestId | Support and trace identifier. | Preserve it in bug reports or support summaries. |
docsUrl | Canonical recovery page. | Read it before recommending a fix. |
fieldErrors | Field, header, parameter, or form-specific issues. | Name the exact affected field and expected shape. |
Recovery decision table
| Situation | Retry? | Better next action |
|---|---|---|
| Invalid JSON, missing required field, bad column name, unsupported model. | No. | Fix the request or job spec, then send a new request. |
| Missing, malformed, revoked, or expired credential. | No. | Replace the credential. |
| Valid principal lacks required scope or role. | No. | Use a credential with the required permission. |
| Cross-organization or missing resource. | No blind retry. | Verify the ID belongs to the authenticated organization. |
| Idempotency fingerprint conflict. | No with the same key. | Reuse the key only for the exact same request, or send a new key. |
| Export not ready. | Poll. | Read the export until it is ready before downloading. |
| Transient intake failure. | Maybe. | Retry only when the upload source still exists and the error is marked retryable. |
| Internal or upstream outage. | After recovery. | Preserve requestId and retry after the service recovers. |
Specific error pages
| Code | Typical next step |
|---|---|
validation_failed | Fix request shape, field values, schema, header, or multipart metadata. |
csv_parse_failed | Fix the uploaded CSV and create or retry the upload. |
transient_intake_failure | Retry when the upload still has a recoverable source. |
upload_source_expired | Create a new upload because the original source is gone. |
Relationship to API errors
This page answers "what should the client do next?" The API errors page answers "what does the HTTP error envelope look like?"
Safety rules
- Never expose API keys, webhook secrets, auth headers, secret hashes, object-store pointers, signed URLs, stack traces, source rows, or generated outputs in an error summary.
- Do not infer whether a resource exists in another organization.
- Do not retry non-retryable validation, authentication, or authorization errors.
- Keep
requestIdvisible for support escalation.