Agent Terminals
Right-click any branch and choose Open agent terminal here. A terminal opens, already in that folder, ready for whichever agent you run — Claude Code, Codex CLI, Gemini CLI. It is your agent and your subscription; there is no API key to paste and no credits to buy.
The terminal and its branch share a colour. With one agent running that is a nicety. With four running in different parts of the vault it is the whole point: you can see at a glance who is working where, and the coral becomes a live dashboard of work in progress rather than a picture of work already done.
Terminals open as pinned tabs, so they stay put instead of splitting your workspace apart. Right-click the branch again to go to its terminal or to stop it.
This needs the community Terminal plugin installed and enabled. Liefwork never spawns a process itself — it asks Terminal to, and if Terminal isn't there it says so and does nothing.
Why run an agent from a branch at all? Because where an agent starts is context. An agent launched at a branch is standing inside the part of the memory that matters for the task, with the meristem above it explaining what this thread is and where it got to. It reads the summary, works, and records what it did in the same place. The map you navigate is the map it grows.