Replies: 1 comment 1 reply
|
just a side note, I forked your latest release and done that feature I mention above |
1 reply
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
idea / problem
I run coding agents across a few always-on boxes plus my laptop. --remote
ssh today gives a full-screen attach to one pane at a time, so watching
several remote hosts means separate SSH sessions or separate herdr windows
per host — no unified view.
requested change
Mount a headless remote herdr session as a new workspace inside a local herdr, so local and remote workspaces sit in one sidebar with native agent-status detection and cold-resume. This federates a whole remote workspace (multiple panes/tabs, agent detection, resize, clipboard) alongside local ones, via a dedicated versioned protocol over the existing SSH bridge to a new remote federation-serve subcommand — version/ capability mismatch falls back to today's attach behavior. Raw PTY byte channels (not the rendered-ANSI attach path), fenced per mount so a remote restart can't leak stale traffic in. Remote agent status relays into the existing detection UI with an origin badge. Trust model: I control both binaries, SSH is the auth boundary, ingestion is still sanitized/bounded.
Scope right now: single remote mount, no warm handoff (cold-resume only), no kitty-graphics over federation, no local-echo prediction.
why you want this
One herdr window shows all my hosts instead of juggling separate SSH sessions or windows per box. I've been running this daily against real remote boxes, fully opt-in, existing --remote ssh path untouched.
Does this fit where herdr is headed? If there's interest I'd like to land it in scope-sized pieces (protocol, remote server, client, UI) rather than one huge PR, once there's an accepted issue to attach the work to.
All reactions