Post

The persistent session solves a pretty real problem with agentic coding tools, losing state when a terminal dies mid-task. The harder part may be multiple agents editing the same repo. How does it handle conflicting changes or merge conflicts between agents?

I’ve been using this daily for last 3 weeks, love it. I tried tmux years ago, but wasnt ready. Now with multi agent stuff, this was great timing.

Ratatui's become the default toolkit for these agent-wrangling TUIs, same base as gitui and bottom. Persistent reconnectable sessions is the real gap though, most people are just running agents in raw tmux panes right now.

The reliability test I’d want is a cold restart, not just reconnect. Can it recover each agent’s command, cwd, repo HEAD, exit state, and identity without attaching to an orphaned process or the wrong worktree? A small recovery receipt would make “persistent” auditable instead of merely visual.