Axel
Agent orchestration, designed for Rex
359 followers
Agent orchestration, designed for Rex
359 followers
Threads and named agents, back in the terminal. Axel orchestrates Claude Code, Codex and OpenCode agents on Rex: durable sessions, status the terminal understands, and your other machines over Tailscale.
This is the 2nd launch from Axel. View more
axel.md
Launching today
Threads and named agents, back in the terminal. Axel orchestrates Claude Code, Codex and OpenCode agents on Rex: durable sessions, status the terminal understands, and your other machines over Tailscale.
Launch Team
ClarifyThe Full-Stack AI CRM that replaces 5 tools
Promoted
absolute fan of @ludovic's work. anyone here using Rex, @mitchellh's latest product? install axel.md.
If I edit a file by hand and then revert an agent turn that went south, what happens to my edit? Does axel only roll back what that turn touched?
@artemkorchunov yep! axel keeps track of each turns using worktrees.
thanks so much @fmerian!
I launched Axel back in February and I've been exploring a ton of different flows since then -
different coding harnesses, skills, Buzz, T3Code, Tailscale, etc.
I recently got my hands on Rex Beta from @mitchellh the Great, it inspired me to rethink my flow from first principles, incorporating all the concepts and features I enjoyed from all the different flows I've tried:
threads, named agents you can summon anytime, remote dispatch, token monitoring, diff views, ability to revert changes from a turn that went south.
Rex is all about little big details that make working from a terminal just great. With this new version re-designed from scratch, I'm focusing on creating a new experience that embraces Rex's philosophy.
I also now have a setup allowing me to live refine my workflow as I work on other projects, the satisfaction I'm getting out of this is insane.
Great tools inspire great work. What a freaking time to be building.
Hope y'all will enjoy axel as much as I do
Named agents you can summon is the part that stuck with me — I run a couple agents in parallel and keeping track of which one is doing what is half the job. Curious how the token monitoring works when a few agents are running at once — is it per thread, or do you get a combined view?