codex-worker

You can run parallel Codex CLI agents by spawning each worker in its own tmux session and git worktree. This isolates disjoint subtasks like multi-file refactors or batch triage, enabling concurrent execution and faster turnaround.

Yes, using tmux and git worktrees isolates parallel coding agents effectively. Each Codex worker operates in a separate tmux session and worktree, ensuring independent task execution without file conflicts across multiple agents.

Yes, you need both tmux and codex installed to orchestrate parallel agents. The tool requires these dependencies to manage multiple agent sessions and coordinate worktrees for isolated execution of independent subtasks.

The best way to parallelize independent coding subtasks is orchestrating multiple Codex CLI agents via tmux. This approach replaces the built-in Task tool when tmux and codex are available, managing separate worktrees for each disjoint subtask.

Use parallel Codex agents instead of the built-in Task tool when your task can be decomposed into independent subtasks and both codex and tmux are available. This enables automatic parallelization for batch triage or parallel fixes.

Separate git worktrees provide isolated execution environments for parallel tmux coding sessions. Each Codex agent works in its own worktree, preventing file conflicts and enabling multiple agents to tackle complex tasks simultaneously.