Launching today
ManyAgents
Run a team of AI coding agents from one Mac window
60 followers
Run a team of AI coding agents from one Mac window
60 followers
An orchestrator splits your work across Claude Code and Codex agents, each in its own git worktree. A reviewer checks the result, a built-in browser lets agents test their own work, and every question lands in one place. Uses the plans you already have. Free for 14 days.
I often have 10+ AI agents running at once: writing code, reviewing changes, running tests and coordinating the work. Across several projects that can triple, and keeping track of who's doing what gets overwhelming.
So I built ManyAgents, a Mac app where each project gets an orchestrator. It hands work to agents in their own git worktrees, has a reviewer check it, and lets them test their changes in a built-in browser beside the chat. Every question lands in one place.
It's free for 14 days. I'm still working out pricing, so I'd love to hear what you'd pay, and what's missing from how you run your agents today.
one place for every agent's question is a nice touch when you're juggling 10+ of them. does each question show which project it came from?
Wow Stu! Never managed too many agents and feel it's worth it. Is there any kind of trial or free tier?
A read-only reviewer on a different provider is the part I like most. Having a second agent check the work before I see it is what made me trust the agents across my projects.
Putting questions from all the workers in one place makes sense. I'd want the project and proposed decision beside each one so I can answer without reopening every thread.
How does the orchestrator merge the worktrees that touch the same lockfile or migrations before the user sees the result ? does conflicrts get serialized or flagged for manual reviews ?
@boutar_badii ManyAgents doesn't merge anything on its own. It flags conflicts early and leaves the merge to an agent or to you. A built-in rule that serializes migrations would make a good item for the next update though.
@stu_green serialization is a good idea. Regarding the lockfile, I would regenerated it from merged package.json, not text-merging it, is that something planned for the next releases ?
@boutar_badii Agreed, regenerating beats text-merging. Today ManyAgents doesn't merge for you, it flags branches that touch the same files and leaves lockfiles out of those warnings. Serialized merges, with the lockfile rebuilt from the merged package.json, are going on the list for an upcoming release. Thanks for the suggestion.
How does the reviewer agent decide a worktree is ready to merge, and what happens when two agents touch the same file? Merge conflicts across parallel branches seem like the hard part.