$ ls blog/

How it works,
in depth.

Deep dives into Agent Interface: tmux, native terminals, workspaces, and the Agent Backbone integration. Written by the people who build it.

featuredhow it works
$ tmux -C
keeps running
$
[date] · [n] min read

Close the window, keep the work: how Agent Interface uses tmux

Quit the app in the middle of a long agent run and nothing stops. Here is why: tmux owns every process, and Agent Interface is only ever a client.

read the post →
how it works layouts.db
[date] · [n] min read

Inside a workspace: how layouts are saved and merged between windows

Libraries, three-way merges and why a conflicting edit asks you to repeat it rather than guessing.
how it works ghostty-vt
[date] · [n] min read

Drawing terminals natively with Ghostty’s core

From tmux control messages to pixels: workers, owned screen snapshots and a UI thread that never waits on I/O.
guides claude + codex
[date] · [n] min read

Running Claude Code and Codex side by side

A workspace layout for a builder and a reviewer agent, with a shell and a server in the panes beside them.
integrations GET /api/agents
[date] · [n] min read

What Agent Interface reads from Agent Backbone

Read-only, local-only and never stale: the rules behind the agents section and the native views.
guides attach -t
[date] · [n] min read

Attaching tmux sessions without taking them over

How attached panes share window sizing with other clients, and how to pick a tmux window without switching anyone else.
releases v[x.y]
[date] · release notes

Agent Interface [version]: what’s new

[Release summary: the headline changes, with links to the full notes on GitHub.]

New posts, as they land.

Follow along by RSS, or watch releases on GitHub.

RSS feed Releases ↗