Scope groups actions so you can handle the group's result. A common pattern puts normal work in a Try scope and configures a Catch scope to run after Try has failed or timed out. The recovery actions must be attached to those statuses. Simply placing a card below a failing action does not make it an error handler.
Run history keeps the failed action visible even when a recovery path completes. That evidence helps support distinguish a recovered incident from a fully successful first attempt. The local fixtures can fail a named connector action, which lets the checks test both the normal and recovery routes. Avoid sending the normal success message from Catch. A handled failure should produce a specific escalation or compensating action rather than hiding that the original work did not complete.