Skip to main content
When the workflow author turns on continue on error, you can choose how each error from your block is returned:
  • An unclassified error has no id. It follows the catch-all outcome, along with every other unclassified error.
  • A classified error has an id, with a corresponding <ErrorOutcome /> in the block configurator. The workflow author can then handle that error on its own outcome.
Both shapes are valid returns from execute and finish.

Unclassified errors

Return an error with no id when you do not expect the workflow author to handle that error on its own outcome:
When the workflow author turns on continue on error, an unclassified error follows the catch-all outcome instead of failing the run.

Classified errors

When a specific error should have its own outcome, add an <ErrorOutcome /> in the configurator, then return that same id in the error payload from execute (or finish) handlers.

Surface the error outcome

ErrorOutcome comes from Workflows.useConfigurator, alongside Outcome. Render one for each error the workflow author should be able to handle:
configurator.tsx
Rendering the <ErrorOutcome /> only lists the error in the workflow author’s error-handling dropdown. It does not add an outcome until the workflow author chooses to handle that error.
label is required on ErrorOutcome. On Outcome it is optional when the block has a single success outcome.

Return the error

From execute or finish, return the same id with an errorMessage and a data payload:
execute.ts
id must match an <ErrorOutcome /> in the configurator. data is that outcome’s payload and must match its schema. Pass null when the schema is null.
retryable is optional, as on an unclassified error. The workflow author decides whether this outcome is handled. When they handle it, the run continues down that error outcome, and downstream steps can use data.

See also