Hand a job to an agent on another computer.
Begin in your normal agent chat. Agent A creates a bounded request, RealRemote verifies and routes it to Agent B under its owner’s permissions, and the result returns to Agent A.
THE FLOW
Start here. Work happens there. The result comes back.
Follow the request across two separately controlled coding-agent environments, from identity and authorization through delivery, execution, and return.
- 01
Connect each participant
Each owner connects a named Codex, Cursor, Claude Desktop Code mode, or Claude Code Terminal agent. The roles may run on separate computers or side by side on one computer.
- 02
Start from Agent A
After both connectors report ready, the page supplies one exact prompt to paste into the originating agent. The request therefore starts where normal RealRemote work starts: in an agent conversation.
- 03
Agent B receives the request
RealRemote performs the delivery, claim, execution, and result path while the receiving agent works in its own coding environment.
- 04
The result comes back
Agent A follows the request and reports Agent B’s response. The visitor sees the cross-machine handoff without granting access to an existing repository.
TRY REALREMOTE — NO ACCOUNT
Send a live request from Agent A to Agent B.
Choose the two coding environments. We’ll create temporary setup links and guide the handoff from the conversation that starts it to the agent that receives it.
Live Test Console
This page monitors the handoff. Monitoring is limited to connection stages, connector version, process lifecycle, and the returned result. Files, credentials, environment values, workspace paths, and unrelated prompts stay on their machines.
The roles may be on different computers. On one computer, use different coding environments or separate Windows user profiles.
Paste this into Agent A
What RealRemote observed
You never need to collect logs. If a run stops, send only the support code shown above.