coding-agent

Run interactive coding agents with PTY-safe terminal execution.

5|Updated Jan 31, 2026
One-click install
npx skills add https://github.com/kcns008/clusterclaw --skill coding-agent-kcns008
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: coding-agent
Source: https://github.com/kcns008/clusterclaw/tree/main/skills/coding-agent
Command: npx skills add https://github.com/kcns008/clusterclaw --skill coding-agent-kcns008

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you run interactive coding agents reliably from the terminal without broken output, hanging sessions, or lost context, especially when you need programmatic control over AI-assisted development work.

Core Features & Use Cases

  • PTY-safe execution: Ensures interactive tools like Codex, Claude Code, OpenCode, and Pi run with a pseudo-terminal so their output and prompts behave correctly.
  • Background session control: Supports long-running tasks with monitoring, log inspection, input submission, and termination when needed.
  • Workflow guidance: Covers one-shot prompts, repository-scoped work, PR review flows, and parallel issue fixing with git worktrees.
  • Use case: Launch a coding agent in a project directory to implement a feature, keep the user updated while it runs, and review the results when the session completes.

Quick Start

Use the coding-agent skill to start your chosen coding agent in the target git repository with PTY enabled and describe the task you want completed.

Frequently Asked Questions about coding-agent

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run coding agents from bash without the terminal session hanging or breaking?

Running coding agents from bash safely requires PTY allocation. This skill ensures interactive tools like Codex and Claude Code execute with a pseudo-terminal, preventing broken output, hanging sessions, and lost context during programmatic control over AI-assisted development.

Can I execute Claude Code or Codex CLI as a background session and monitor the output?

Yes, you can execute Claude Code or Codex CLI as a background session. This skill supports long-running tasks with monitoring, log inspection, input submission, and termination when needed for safe bash automation of coding agents.

What is the best way to fix multiple GitHub issues in parallel using coding agents?

The best way to fix multiple issues in parallel is using git worktrees. This skill provides workflow guidance for parallel issue fixing, allowing coding agents to work on isolated repository branches simultaneously without file conflicts.

Do I need trusted git workdirs to automate PR review workflows with Codex?

Yes, you need trusted git workdirs to automate PR review workflows with Codex. This skill requires explicitly defined trusted workdirectories and explicit stdin handling to ensure safe automation and prevent unintended file modifications.

Why does my interactive coding agent lose context when launched from a bash script?

Interactive coding agents lose context in bash scripts due to missing PTY allocation and improper stdin handling. This skill solves context loss by allocating a pseudo-terminal and managing explicit stdin input for safe programmatic execution.