A robust orchestration layer for running interactive AI coding agents like Codex, Claude Code, and Gemini via a controlled terminal environment.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install multi-coding-agent
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 multi-coding-agent 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 Coding Agent Skill?
The Coding Agent skill provides a sophisticated framework for managing interactive terminal-based AI agents. It leverages pseudo-terminal (PTY) support to ensure that agents like Codex, Claude Code, and Pi Coding Agent function correctly without hanging or broken output. By utilizing background processes and session monitoring, this skill allows developers to automate complex coding tasks, perform batch pull request reviews, and manage parallel issue fixing across multiple git worktrees. This is an essential component for developers looking to integrate advanced AI capabilities into their local development environment via Openclaw Skills.
Built to be bash-first, this skill prioritizes control and visibility. It allows for one-shot command execution or long-running background sessions that can be polled, logged, and interacted with in real-time. Whether you are building new features or refactoring legacy code, the Coding Agent skill ensures your AI tools have the right environment to succeed while keeping your primary workspace clean.
Coding Agent Skill Use Cases
- Executing one-shot coding prompts or interactive chat sessions in isolated environments using Openclaw Skills.
- Automating batch pull request reviews across multiple repositories using parallel agent instances.
- Orchestrating complex refactors or full-stack feature builds in the background while monitoring progress via session logs.
- Fixing multiple GitHub issues simultaneously by deploying agents into dedicated git worktrees.
- Implementing a model fallback strategy to switch between Codex, Claude, and Gemini when hitting API limits.
How Coding Agent Skill Works
- The user initiates a command via the bash tool, ensuring the pty:true flag is set to accommodate interactive terminal requirements.
- The skill launches the specified coding agent (e.g., Codex, Claude, or Pi) within a designated working directory to maintain context isolation.
- For long-running tasks, the agent runs in a background session, returning a unique sessionId for tracking.
- Developers use process tools to monitor logs, check status via the poll action, or send manual input using write or submit actions.
- Upon completion, the agent can optionally trigger a wake notification to signal that the task is finished without waiting for manual checks.
Coding Agent Skill Setup
Ensure you have at least one supported coding agent installed on your system. The skill supports Codex, Claude Code, Gemini CLI, and Pi Coding Agent.
# Example: Install Pi Coding Agent
npm install -g @mariozechner/pi-coding-agent
# Basic execution pattern (Always use PTY!)
bash pty:true workdir:~/your-project command:"codex exec 'Refactor the API'"
Note: Codex requires a git repository to function. For scratch work, initialize a temporary repo using git init.
Coding Agent Skill Data Schema & Taxonomy
The Coding Agent skill manages process metadata and file system context to ensure reliable execution.
| Attribute | Description |
|---|---|
| sessionId | Unique identifier for tracking background coding processes |
| workdir | The isolated project directory providing context to the agent |
| pty | Boolean flag (must be true) to allocate a pseudo-terminal |
| command | The specific CLI command string executed for the AI agent |
| logs | The captured stdout/stderr from the active agent session |
Coding Agent Skill Advanced Features
- Parallel Army: Deploy multiple coding agents simultaneously for large-scale code analysis or batch PR reviews.
- Git Worktree Integration: Use isolated branch management to fix multiple issues in parallel without workspace conflicts.
- tmux Orchestration: Seamlessly hand off background tasks to tmux for session persistence and interactive debugging.
- Auto-Notify on Completion: Utilize the moltbot gateway wake trigger to receive immediate alerts when background agents finish their tasks.
- Context Forking: Transfer captured terminal context from one agent to another to facilitate multi-stage planning and execution workflows.
SKILL.md
Loading
Related Openclaw Skills
MinerU PDF Parser
A local, CPU-based PDF parser that transforms complex documents into AI-ready Markdown and JSON formats.
lobster-jobs
A specialized migration tool for converting OpenClaw cron jobs into deterministic Lobster workflows featuring approval gates and resumable execution.
Klutch Programmable Finance
An advanced financial integration that enables AI agents and developers to interact with Klutch programmable credit cards through a secure GraphQL API.
Stealth Browser
An advanced browser automation skill designed to bypass aggressive anti-bot protections like Cloudflare and Datadome using stealth-patched browsers.
Oura Analytics
A comprehensive integration for Oura Ring that automates health data retrieval, trend analysis, and recovery alerting.
PyMuPDF PDF Parsing
A lightning-fast local tool for converting PDF documents into Markdown and JSON formats with support for image and table extraction.
Featured*
