Currently, errors caused by any part of the Agent's configuration (e.g. insufficient API credits, invalid API key) are not propagated to the Sandbox during testing. This means that the Agent just doesn't return anything, which can be hard for the end user to debug. Instead, these errors should be shared in the Sandbox so that the user can react appropriately.
Description
Currently, errors caused by any part of the Agent's configuration (e.g. insufficient API credits, invalid API key) are not propagated to the Sandbox during testing. This means that the Agent just doesn't return anything, which can be hard for the end user to debug. Instead, these errors should be shared in the Sandbox so that the user can react appropriately.