harness

Harness

Agent Team & Skill Architect — Works with both Claude Code and OpenAI Codex CLI

English | 한국어 | 日本語

Forked from revfactory/harness. This version supports both Claude Code and Codex CLI from a single repository.

A meta-skill that designs domain-specific agent teams, defines specialized agents, and generates the skills they use.

Installation

Claude Code

# Via marketplace
/plugin marketplace add sumniy/codex-harness
/plugin install harness@harness

# Or direct copy
cp -r skills/claude-code/harness ~/.claude/skills/harness

OpenAI Codex CLI

# Via $skill-installer
$skill-installer install https://github.com/sumniy/codex-harness/tree/main/skills/codex/harness

# Or direct copy
cp -r skills/codex/harness ~/.agents/skills/harness

Repository Structure

harness/
├── .claude-plugin/                    # Claude Code plugin manifest
│   ├── plugin.json
│   └── marketplace.json
├── .codex-plugin/                     # Codex CLI plugin manifest
│   ├── plugin.json
│   └── marketplace.json
│
├── skills/
│   ├── claude-code/                   # ← Claude Code version (original)
│   │   └── harness/
│   │       ├── SKILL.md
│   │       └── references/
│   │           ├── agent-design-patterns.md
│   │           ├── orchestrator-template.md
│   │           ├── team-examples.md
│   │           ├── skill-writing-guide.md
│   │           ├── skill-testing-guide.md
│   │           └── qa-agent-guide.md
│   │
│   └── codex/                         # ← Codex CLI version (adapted)
│       └── harness/
│           ├── SKILL.md
│           ├── agents/openai.yaml
│           └── references/
│               ├── agent-design-patterns.md
│               ├── orchestrator-template.md
│               ├── team-examples.md
│               ├── skill-writing-guide.md
│               ├── skill-testing-guide.md
│               └── qa-agent-guide.md
│
├── README.md
├── README_KO.md
└── README_JA.md

Platform Comparison

	Claude Code	Codex CLI
Skill path	skills/claude-code/harness/	skills/codex/harness/
Agent defs	.claude/agents/{name}.md	.codex/agents/{name}.toml
Skill defs	.claude/skills/{name}/skill.md	.agents/skills/{name}/SKILL.md
Execution mode	Agent Teams (default) + Subagents	Subagents only (parallel via max_threads)
Top model	model: "opus"	model = "o3" / gpt-5.4
Install cmd	/plugin install	$skill-installer install <url>
Invoke	/harness or auto-trigger	$harness or auto-trigger
Project docs	CLAUDE.md	AGENTS.md

Key Features

• 6 Architecture Patterns — Pipeline, Fan-out/Fan-in, Expert Pool, Producer-Reviewer, Supervisor, Hierarchical Delegation
• Skill Generation — Progressive Disclosure for efficient context management
• Orchestration — Inter-agent data passing, error handling, coordination protocols
• Validation — Trigger verification, dry-run testing, with-skill vs without-skill comparison

6-Phase Workflow

Phase 1: Domain Analysis
Phase 2: Team Architecture Design
Phase 3: Agent Definition Generation
Phase 4: Skill Generation
Phase 5: Integration & Orchestration
Phase 6: Validation & Testing

Usage

Claude Code:

하네스 구성해줘
Build a harness for this project

Codex CLI:

$harness
하네스 구성해줘
Build a harness for this project

Credits

• Original: revfactory/harness by @revfactory
• Research: revfactory/claude-code-harness — +60% quality improvement
• Harness 100: revfactory/harness-100 — 100 pre-built harnesses across 10 domains

License

Apache 2.0