Codex Orchestrator for Openclaw

A specialized workflow for supervising, monitoring, and controlling background Codex AI agent sessions to ensure task completion.

microcarft
v1.0.0
Feb 6, 2026
1
4.8k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install codex-orchestrator

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install codex-orchestrator using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is Codex Orchestrator?

Codex Orchestrator provides a robust framework for managing long-running AI coding tasks without blocking your main development thread. By utilizing background PTY sessions, this skill allows developers to delegate complex coding operations to the Codex agent while maintaining full supervisory control. It bridges the gap between autonomous execution and manual oversight, ensuring that background processes remain productive and recoverable.

Integrating this skill into your environment enhances the reliability of Openclaw Skills by providing mechanisms to handle interactive prompts, detect hung processes, and manage session lifecycles. Whether you are refactoring large codebases or generating complex features, this orchestrator keeps your AI workflows transparent and manageable.

Codex Orchestrator Use Cases

  • Executing long-running coding tasks in the background without locking the terminal
  • Monitoring the real-time progress of autonomous AI agents during complex refactoring
  • Handling interactive CLI prompts that would otherwise stall a background process
  • Recovering or resuming AI coding sessions after interruptions or process failures
  • Periodic health checks on autonomous agents to detect hallucinations or infinite loops

How Codex Orchestrator Works

  1. Launch the Codex agent in a background PTY session using specialized bash commands to capture a session ID.
  2. Monitor progress by periodically retrieving log snapshots to distinguish between active work and blocked states.
  3. Intervene when necessary by submitting input data to handle interactive prompts or killing unresponsive sessions.
  4. Apply Standard Operating Procedures (SOPs) to diagnose and resolve stuck agents or resume interrupted work.
  5. Summarize milestones and notify the user upon task completion or significant progress.

Codex Orchestrator Setup

To begin using this skill within your Openclaw Skills environment, ensure the Codex CLI is installed. Launch a background session with the following command:

bash pty:true workdir:<target_dir> background:true command:"codex exec --full-auto '<PROMPT>'"

To find a lost session ID, use:

process action:list

Codex Orchestrator Data Schema & Taxonomy

Component Description
sessionId Unique identifier for the background PTY session
PTY Buffer Ephemeral storage for real-time logs and process output
Snapshots Periodic log captures via process action:log with a 2KB limit
Persistence Optional redirection to codex.log for long-term audit trails

Codex Orchestrator Advanced Features

  • Automatic diagnosis of stuck processes via the 'Stuck Agent' protocol
  • Remote intervention to answer interactive CLI questions via process action:submit
  • Session resumption capabilities using codex resume to pick up context after crashes
  • Non-blocking background execution allowing for multi-agent orchestration
  • Resource management through targeted session termination and process cleanup

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*