Coding Agent for Openclaw

A background-first skill to programmatically control AI coding agents like Codex, Claude Code, and Pi for automated development workflows.

thiagoruss0
v1.0.0
Feb 7, 2026
0
0
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install coding-agentagb2

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 coding-agentagb2 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?

The Coding Agent skill provides a robust framework for running AI-powered development tools in the background, allowing for non-interactive and parallelized coding work. By utilizing a workdir-centric pattern, it ensures that agents like Codex CLI, Claude Code, and OpenCode operate within focused, isolated environments, preventing them from accessing unrelated system files. This skill is a core component of Openclaw Skills, designed to transform how developers handle repetitive tasks like building components or reviewing pull requests.

Whether you need a sandbox for experimental features or a high-speed YOLO mode for rapid prototyping, this skill offers the programmatic hooks necessary to manage long-running AI processes. It integrates seamlessly with process management commands to monitor logs, poll for status, and send inputs to agents when human intervention is required.

Coding Agent Use Cases

  • Running high-speed feature generation using the --yolo or --full-auto flags in Codex.
  • Executing parallel pull request reviews by deploying a fleet of AI agents across multiple git worktrees.
  • Fixing multiple repository issues simultaneously without blocking your primary development terminal.
  • Automating routine code refactoring or documentation tasks across large-scale projects using Openclaw Skills.

How Coding Agent Works

  1. Define a focused workdir or temporary scratch directory to isolate the agent's file access.
  2. Execute the desired coding agent command using the bash background mode to initiate a non-blocking session.
  3. Capture the returned sessionId to track the lifecycle of the specific coding task.
  4. Monitor the implementation progress using the process log and poll actions.
  5. Handle interactive prompts by sending data directly to the session if the agent requires clarification.
  6. Finalize the task by reviewing the generated code, committing changes, or killing the process if complete.

Coding Agent Setup

Before using this skill, ensure that OpenClawCLI is installed. You must also have your preferred coding agents installed on your system.

# Example: Installing Pi Coding Agent
npm install -g @mariozechner/pi-coding-agent

# Configure your Codex model in ~/.codex/config.toml
# Ensure the bin (claude, codex, opencode, or pi) is in your PATH

Coding Agent Data Schema & Taxonomy

The skill organizes its operations through session management and directory isolation. The following table describes how data and metadata are handled within Openclaw Skills:

Property Description
sessionId Unique identifier for tracking the background agent process.
workdir The designated filesystem path where the agent is localized.
command The specific CLI string passed to the agent (e.g., codex, claude).
background Boolean flag indicating the process runs as a background task.
logs Captured stdout and stderr used for monitoring the agent's logic.

Coding Agent Advanced Features

  • Parallel Army Deployment: Launch multiple Codex instances simultaneously to process a batch of PRs or issues.
  • Git Worktree Integration: Use isolated branches for each agent to prevent conflicts during parallel code modification.
  • tmux Skill Synergy: Switch to tmux mode for interactive coding sessions that require a persistent TTY while still benefiting from Openclaw Skills.
  • Safety Guardrails: Built-in protections to prevent agents from running in sensitive system directories or the live project root.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*