An autonomous orchestration skill for managing multi-agent development tasks using tmux, Codex, and intelligent dependency analysis.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install ec-task-orchestrator
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install ec-task-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 Task Orchestrator?
The Task Orchestrator is a sophisticated addition to the Openclaw Skills ecosystem, designed to manage complex, multi-step engineering projects through autonomous agent coordination. By utilizing tmux sessions and git worktrees, it allows multiple AI agents to work on separate issues simultaneously without file conflicts. This skill is essential for teams looking to scale their AI-driven development by enforcing senior-level engineering principles and structured project management.
With Openclaw Skills like this one, you can transform a list of GitHub issues into a structured execution plan. The orchestrator analyzes file dependencies to decide which tasks can run in parallel and which must be serialized, ensuring your codebase remains stable while maximizing throughput. It acts as the brain of your AI coding fleet, managing the lifecycle of every task from initial analysis to the final pull request.
Task Orchestrator Use Cases
- Parallel processing of multiple GitHub issues across independent worktrees
- Large-scale refactors requiring coordinated changes across multiple modules
- Autonomous CI/CD lifecycle management from issue analysis to PR creation
- Managing high-reasoning AI agent sessions with automated self-healing and monitoring
- Orchestrating complex software builds with multi-phase dependency gates
How Task Orchestrator Works
- The orchestrator initializes a JSON manifest to track project status, dependencies, and file mappings.
- It fetches open issues via the GitHub CLI and analyzes them to identify file-level conflicts or explicit dependency requirements.
- Isolated git worktrees are created for each task to ensure a clean, branch-based environment for the AI agents.
- Tasks are launched in dedicated tmux sessions, using Codex in yolo mode for autonomous execution.
- A heartbeat monitor periodically polls all active sessions, checking for completion, errors, or stalled prompts.
- Upon successful completion, the system automatically pushes the changes and opens a detailed pull request.
Task Orchestrator Setup
To begin using this entry from Openclaw Skills, set up your working directory and initialize the manifest:
# 1. Create working directory
WORKDIR="${TMPDIR:-/tmp}/orchestrator-$(date +%s)"
mkdir -p "$WORKDIR"
# 2. Clone repository and initialize tmux socket
git clone https://github.com/OWNER/REPO.git "$WORKDIR/repo"
SOCKET="$WORKDIR/orchestrator.sock"
# 3. Analyze GitHub issues to generate the task list
gh issue list --repo OWNER/REPO --state open --json number,title,body,labels > issues.json
Task Orchestrator Data Schema & Taxonomy
The Task Orchestrator organizes its lifecycle through a manifest.json file. Below are the core metadata components and task status definitions:
| Status | Meaning |
|---|---|
pending |
Task queued for execution |
blocked |
Waiting for a dependency to complete |
running |
Active Codex session in tmux |
stuck |
Agent is waiting for input or no progress detected |
complete |
Task logic finished successfully |
pr_open |
Pull Request has been created on GitHub |
Manifest Structure Highlights:
project: Unique name for the orchestration runphases: Array of task groups with dependency gatesdependsOn: Array of task IDs that must finish before execution
Task Orchestrator Advanced Features
- Dependency-Aware Phase Gating: Automatically pauses execution of downstream tasks until critical dependencies are met.
- Self-Healing Heartbeat: A cron-based monitor that detects stalled AI sessions and automatically sends 'y' responses or restarts sessions with failure context.
- Parallel Execution Batches: Intelligent grouping of tasks that touch different files to maximize CPU and model throughput.
- Isolated Worktree Environments: Prevents merge conflicts during development by giving every agent its own git worktree.
- Auto-Recovery with Reasoned Retries: When an error occurs, the orchestrator captures the last 100 lines of logs and feeds them back into the model for a corrective second attempt.
SKILL.md
Loading
Related Openclaw Skills
BMAD Orchestrator
A comprehensive orchestration skill that manages the 4-phase BMAD Method across OpenClaw and Claude Code for agile, AI-driven development.
Henteplan - Norwegian Waste Collection
A specialized tool for Norwegian households to query waste collection schedules and set up automated bin reminders.
Kalshi Prediction Markets
A read-only Kalshi integration for browsing prediction markets and analyzing trade opportunities through Openclaw Skills.
Corporate Credit Memo Generator
An automated financial analysis tool that transforms raw corporate reports into comprehensive, bank-ready credit application memoranda.
OpenClaw Memory Enhancer
An edge-optimized RAG memory system that provides AI agents with persistent, semantic long-term memory using minimal system resources.
Feishu Deep Research
An automated research agent that conducts deep data gathering and generates structured reports directly in Feishu cloud space.
Featured*
