The refusals and empty screens that mean something specific, and what each one is actually telling you.
Troubleshooting
Most confusing screens in this app are one of a handful of specific situations. Each of these means something different, and the fixes have nothing in common.
Screens are locked and say to select a project
No project is selected. Use the picker at the top of the sidebar, or create one on Organization → Projects. A brand-new organization has none, which is why organization-wide screens such as Organization, Approvals and Access requests stay reachable.
Creating a project needs one of the organization-administration permissions, the same set that puts
Organization in the sidebar at all. If you do not have them, ask an administrator to create a project
and give you access to it. The guide itself is public at xauto.dev/docs/; it is no longer an in-app
screen.
A screen shows nothing, or far less than you expect
In rough order of likelihood:
The wrong environment is selected. The selector is app-wide, not per screen.
You have an environment assignment that does not include this one. An unassigned environment of a scoped project is unreachable, including for reading — see Roles and access. This is the one that looks most like data having vanished.
The connection's credential is scoped narrowly. A partial read is reported in an amber notice naming what could not be read, rather than failing the whole screen.
Nothing has been imported yet. Tracking is explicit — see Tracking resources.
"Test connection" fails
The stored credential no longer authenticates: rotated, expired, or the account it belongs to changed. Register the new credential on Connectors — it is written straight to the secret store and never comes back out through the API.
A control is greyed out
Hover it. Every disabled control in this app carries a reason. The usual three are: your role does not include it, your plan does not include it, or the resource is mid-update and the provider will refuse a second change until it finishes.
An action was refused and named a plan
An organization limit or an ungated capability. Organization → Plan & billing names what fits. If your organization is on a negotiated plan, the refusal points at your account manager instead, because the self-serve checkout cannot express your contract.
A background job failed
Open Jobs and read the error — it is the provider's own message, kept verbatim, and it is usually the useful part. The commonest causes are a credential that no longer authenticates and a permission the credential lacks for that one call.
You expected a notification and got nothing
Every notification type is muted by default and opted into per operator. Open the bell's settings and enable the types you want. Alert email is separate again, per project, and configured by an admin.
The app looks out of date
Open About & updates in the account menu. It reports the build this browser is running, checks for a newer one, and offers a reload when one is downloaded.
An installed app on a phone is resumed far more often than it is loaded, which is the usual reason a device sits on an old build. A failed check is reported as a failed check — being offline is the ordinary reason, and "we could not ask" is not the same answer as "there is nothing new".
Something is genuinely wrong
The Activity feed records every mutating action with who did it and when, including actions taken by support staff while assisting your organization — those are attributed to both the staff member and the operator identity they were acting as. That is the record to read when the question is "what changed".