skills.rest
JOHNNYWHITEMIKE
JOHNNYWHITEMIKECOMMUNITY
Explore all skills by this author →
GitHub

coding-agent

Delegate multi-step coding tasks to external agents for builds, refactors, and PR reviews.

Updated Apr 8, 2026
Search Tags:#refactoring#code generation#workspace isolation#coding agents#pull request review#background sessions#terminal PTY
One-click install
npx skills add https://github.com/JOHNNYWHITEMIKE/openclaw --skill coding-agent-johnnywhitemike
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: coding-agent
Source: https://github.com/JOHNNYWHITEMIKE/openclaw/tree/main/openclaw/.openclaw/openclaw/skills/coding-agent
Command: npx skills add https://github.com/JOHNNYWHITEMIKE/openclaw --skill coding-agent-johnnywhitemike

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coding agents help you move faster on real development work by exploring files, generating multi-step code changes, and handling iterative refactors without you manually orchestrating every edit.

Core Features & Use Cases

  • Delegated multi-step coding: Spawn Codex, Claude Code, or Pi agents to build new features/apps, refactor large codebases, and perform PR review workflows.
  • Safe workspace targeting: Run agents in a specific working directory (including temp/git-cloned review directories) to reduce accidental edits to sensitive locations.
  • Interactive terminal compatibility: Uses PTY where required (Codex/Pi/OpenCode) and the correct Claude Code flags to avoid hangs and broken output.
  • Background execution with monitoring: Start longer sessions in the background and track progress via process actions (list, poll, log, etc.).

Quick Start

Ask the agent to implement your change by running it in the target project directory with a PTY for Codex (or the correct Claude Code flags for Claude Code).

Frequently Asked Questions about coding-agent

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

FAQPage Schema
How do I delegate multi-step coding tasks like refactoring to an AI agent?▼

You can delegate multi-step coding tasks to AI coding agents by spawning external sessions for iterative refactoring and code generation. The agent explores files and executes multi-step edits within a specific working directory.

Can I run coding agents in a specific workspace to prevent accidental file edits?▼

Yes, you can run coding agents in a specific working directory to prevent accidental edits. You can use temporary git repositories for scratch work or git worktrees and temp directories for safe pull request review.

What's the best way to review a pull request using an AI coding agent?▼

The best way to review a pull request is to clone the repository into a temporary directory and run the coding agent there. This isolates the PR review workflow and ensures safe execution without modifying your main codebase.

Why does Claude Code hang in the terminal and do I need a PTY for coding agents?▼

Claude Code requires non-PTY execution with bypassPermissions and print flags to avoid hangs and broken output. In contrast, you do need a PTY for interactive terminal compatibility when running Codex, Pi, or OpenCode agents.

How do I monitor a background coding agent session for a long codebase refactor?▼

You can monitor a background coding agent session by polling its progress via process actions. Start longer coding sessions in the background and track their status using list, poll, and log commands to observe ongoing refactoring.

Related Skills

Explore More Software Engineering Skills→
dakshrawat298-gifdakshrawat298-gif

coding-agent

Manage coding agents in terminal sessions across git worktrees.

Community
Advanced
lllooollpplllooollpp
39

coding-agent

Execute AI coding agents in background shell processes with PTY support.

Community
Advanced
doitiandoitian
14

using-git-worktrees

Create isolated Git worktrees for feature development with automatic setup.

Community
Advanced
TitoPrauseeTitoPrausee

claude-code

Delegate coding tasks to the Claude Code CLI with permission handling.

Community
Advanced