Can I resume a worker after stopping it?

Yes, the toolkit includes 'stop' and 'resume' commands that preserve the state and log history, allowing you to pick up exactly where the worker left off.

What is the advantage of using tmux for Codex workers?

It enables persistent, long-running AI processes that can be monitored, detached, and reattached without losing context or interrupting the task.

How does this skill prevent log cross-talk?

Each worker is assigned a unique ID and an isolated CODEX_HOME directory, ensuring that logs and history remain specific to that individual session.

What are parent and child workers?

The skill allows you to 'spawn' child workers linked to a specific parent, creating a structured tree of AI agents for complex, multi-layered workflows.

How does the 'ask' functionality work?

It uses 'tmux send-keys' to inject text into the worker pane and then polls the JSONL session logs until the assistant's response is detected and returned.

Tmux Workflow Manager

Name: Tmux Workflow Manager
Author: kunhuang12345

bykunhuang12345

生产力与工作流

Orchestrates multiple long-running Codex CLI workers through tmux sessions to enable isolated, concurrent AI task management.

关于

The Tmux Workflow Manager is a comprehensive toolkit designed to drive persistent Codex CLI instances from external processes like the Claude CLI. By leveraging tmux, it provides a robust framework for starting, pausing, and resuming AI workers while maintaining complete environment isolation via unique state directories. This skill is ideal for developers needing to manage complex, hierarchical agent structures (parent-child workers) or those who require reliable log-based communication between CLI tools and AI models without cross-talk or session interference.

主要功能

Hierarchical worker management supporting parent-child agent trees
Stateful session persistence allowing for stopping and resuming workers
Isolated worker environments with per-worker CODEX_HOME directories
Real-time prompt injection and automated log polling for replies
Built-in health checks and diagnostic tools for monitoring worker status
0 GitHub stars

使用场景

Running multiple parallel AI coding agents for complex repository refactoring
Automating long-running CLI-based AI tasks that require persistent session state
Creating structured, multi-agent workflows where a lead agent spawns sub-workers

关于

主要功能

Hierarchical worker management supporting parent-child agent trees
Stateful session persistence allowing for stopping and resuming workers
Isolated worker environments with per-worker CODEX_HOME directories
Real-time prompt injection and automated log polling for replies
Built-in health checks and diagnostic tools for monitoring worker status
0 GitHub stars

使用场景

Running multiple parallel AI coding agents for complex repository refactoring
Automating long-running CLI-based AI tasks that require persistent session state
Creating structured, multi-agent workflows where a lead agent spawns sub-workers