codex-worker

To run parallel Codex agents in tmux, this Skill spawns and orchestrates multiple Codex CLI agents, each in its own tmux session and optional git worktree, to execute decomposed subtasks concurrently with isolated environments and traceable progress.

The best way to orchestrate multi-agent Codex workers for batch refactors is distributing work automatically across isolated tmux sessions and git worktrees, enabling concurrent multi-file refactors with reproducible results and reduced wait times.

Yes, you can use git worktrees to isolate parallel Codex CLI agents. This Skill implements per-task worktrees and isolated environments within tmux sessions, ensuring reproducible results when handling decomposed subtasks like batch triage and parallel fixes.

Yes, tmux supports running multiple Codex agents by spawning each agent in its own tmux session. This approach handles independent workstreams in parallel, applying isolated environments and traceable progress for scalable automation of decomposed tasks.

You should use parallel Codex agents instead of a single agent when your tasks can be split into independent workstreams such as batch triage, parallel fixes, and multi-file refactors. This reduces wait times and improves throughput via concurrent execution.

Spawning multiple Codex workers in tmux improves automation throughput by executing decomposed subtasks concurrently rather than sequentially. Each worker operates in an isolated tmux session and git worktree, reducing overall wait times for batch triage and refactors.