Skip to main content
The Everabyte backup agent is a first-party desktop application for Windows, macOS, and Linux. It backs up a machine to Everabyte storage on a schedule. This page describes it from a developer’s perspective. It is not a guide to building your own agent against platform internals.

​
What it does

  • Enrolls a device with an Everabyte account.
  • Runs scheduled or manual backup jobs from selected paths.
  • Uploads files through the same encrypted storage pipeline.
  • Records runs, including partial runs.
  • Supports restore from a chosen restore point.

​
Authentication

The agent uses a device-bound API key. This is the one documented exception to the company-only rule for the REST surface.
A device-bound backup key is not a substitute for a developer API key. Do not extract one from the agent and use it for your own integration. Blocked devices receive 403 DEVICE_BLOCKED.

​
Partial runs

A run that could not read some files completes as partial, not success. The paths that failed appear in an error sample, and the outcome is visible in the console and notifications. This is deliberate. A backup that silently skips an important locked file is worse than a backup that reports a partial result.

​
Open and locked files

Windows Volume Shadow Copy (VSS) is not implemented. Files locked by another process, such as Outlook PST files or open databases, may be skipped. Practical mitigations:
  • close applications that hold locks before the backup window,
  • schedule backups outside working hours,
  • exclude directories that are always locked, explicitly, so a partial run is expected rather than surprising,
  • verify after runs that critical paths were included.

​
Restore points

Runs can produce restore points, subject to a retention count per job. The most recent restore point is always kept. Restore verification is server-side: a sample from a restore point is decrypted and checked. That demonstrates usability rather than mere presence. Backup jobs can also schedule verification tests that the agent performs locally: it restores a sampled set of files into an isolated directory and reports the outcome.

​
Limits

Because objects live in one region, the agent is a backup against deletion and corruption, not against losing a region. Keep an independent copy if regional loss matters to you.

​
Agent versus API integration

The agent is a product, not an SDK. Treat it as a consumer of the platform, not as an interface you build against.

Upload files

The pipeline the agent writes through.

Versions and restore

Restore semantics.

Data residency

Why regional loss matters.

Integration overview

Other paths.