FILES
Your files keep their structure.
Pull results, logs, and assets with their paths intact.
See sample files
results/metrics.csvrun.log
FILE HANDOFFS BETWEEN YOUR MACHINES
Async file transfer.
Your other machine is offline.
Send large files from your terminal while the receiving machine is offline. Keep files and handoff notes in one Bundle, with workspace history for the next step.
01 / THE BUNDLE
A Bundle carries your files
and the context to continue.
A required note explains what the files are for and what to do next. Handoff history stays in your workspace for reference.
CONTEXT
The next step stays clear.
Keep the purpose and next step with the files, so you can pick up where you left off.
“Run B, seed 42. Compare with the baseline before rerunning.”
HISTORY
Each handoff leaves a record.
Look back at what was sent, by whom, from which device, and when.
See sample history
- 09:10Published · workstation
- 18:30Download requested · laptop
Illustrative Bundle · Files arrive with their note. History remains in the workspace.
02 / FOR AGENTS & WORKFLOWS
Agents use Relay through
CLI + Skills.
CLI + JSON for scripts. Relay Skills guide agents through handoffs. Invite teammates when you need a shared workspace.
CLI fits your scripts.
Push and pull Bundles. Read the results as JSON.
CLI setup & demoSkills guide your agent.
A Relay Skill guides compatible agents through sending, receiving, and history.
Explore the Relay SkillA workspace connects your team.
Invite teammates with their own accounts. The owner manages workspace access.
Explore workspace collaborationWant in when we open up?
The public beta is not open yet. Leave your email to receive the opening notice.
You can also contact us at[email protected]
Personal device handoffs · Team sharing in testing · No end-to-end encryption
WHEN THE BETA OPENS
- a working service address
- the install command for your platform
- a quick-start guide for your first handoff
Frequently asked questions
Getting started4
How is this different from rsync, Git LFS, or a cloud drive?
All three assume something Relay does not need. rsync and scp need both ends reachable at the same time, and they need you to know the destination path. Git LFS keeps large files tied to commits, so a run's output becomes permanent repository history. Cloud drives assume the same folders exist on both machines for the long term. Relay is built for a one-shot handoff where nobody has to be waiting on the other end.
Which platforms are supported?
Windows, Linux, and macOS, wherever Node.js 20.x runs. There is no mobile client today.
What if the machine does not have Node.js?
It has to be installed first — Node.js 20.x is the only requirement. That is the one prerequisite the setup script cannot skip for you.
Do the CLI and the web UI share one account?
Yes. Both are entry points to the same account, devices, and workspaces, so a Bundle pushed from one side is visible from the other.
File handoffs6
Can I retract a transfer after sending it?
Owners and FULL members can move a Bundle to trash, where it can be restored for 30 days. PUSH members cannot delete it. A copy already downloaded stays on the receiving machine, and existing download links may remain valid for up to an hour.
Will pull overwrite files I already have?
No. Pull refuses to overwrite existing files, and it rebuilds the directory structure as it was. Nothing is replaced silently.
What happens if the connection drops mid-transfer?
Large files are uploaded in parts. A resumed transfer only sends the parts that are missing, rather than starting the file again.
How does the receiving machine know something arrived?
In the CLI, run relay bundle inbox to check unread Bundles for this device, then pull what you need. The web UI can show live updates while connected; Relay does not automatically download files to your machines.
What if two devices pull the same Bundle?
Each device tracks its own read state, so both can pull it independently. Neither one sees the other's local copy.
Are permissions and symlinks preserved?
Directory structure and relative paths are rebuilt faithfully. Empty folders are not transferred, because only files are tracked.
Privacy, storage & access7
Is there end-to-end encryption?
Not yet. Traffic is HTTPS and objects are encrypted at rest, but we can technically access stored objects. Do not use the alpha for highly sensitive data.
What does Relay store about my files?
To rebuild your file tree on the receiving side, Relay has to store file names, relative paths, and the note you write. The note is kept as plain text: it is not incidental, it is what tells the other side why the files arrived.
Can I self-host it?
Not yet, and there is no date for it. The current alpha is managed cloud only.
Can I hand files to a colleague, or only to my own machines?
Both. Your own devices use one account. In the collaboration test environment, you can also invite colleagues into a workspace with their own accounts. The owner manages access; public availability will be announced separately.
How much can I store?
During the beta, storage is free and pricing will be announced when the service opens. Capacity is counted across the workspaces you own; teammates' uploads use the owner's allowance. Storage limits depend on the service configuration. The current per-file limit is 5 GiB.
What happens to my data if the service shuts down?
Workspace owners can use relay export to save workspace records as JSON. It does not contain the file bytes, so download the files separately. Members can pull files they can access, but workspace export is owner-only.
When does the beta open?
Not yet — we are still testing with a small group. Request access above and you will get the service address when it opens.