AI Skills

<p align="center"> <img src="docs/images/claude-harness-logo-with-text.png" alt="Claude Harness" width="400"> </p> <p align="center"> <strong>Plan. Work. Review. Ship.</strong><br> <em>Turn Claude Code into a disciplined development partner.</em> </p> <p align="center"> <a href="VERSION"><img src="https://img.shields.io/badge/version-3.3.1-blue.svg" alt="Version"></a> <a href="LICENSE.md"><img src="https://img.shields.io/badge/License-MIT-green.svg" alt="License"></a> <a href="docs/CLAUDE_CODE_COMPATIBILITY.md"><img src="https://img.shields.io/badge/Claude_Code-v2.1+-purple.svg" alt="Claude Code"></a> <img src="https://img.shields.io/badge/Skills-5_Verbs-orange.svg" alt="Skills"> <img src="https://img.shields.io/badge/Core-TypeScript-blue.svg" alt="TypeScript Core"> </p> <p align="center"> English | <a href="README_ja.md">日本語</a> </p>

Why Harness?

Claude Code is powerful—but without structure, it can be unpredictable.

Three commands. One workflow. Production-ready code.

graph LR
    A[Your Idea] --> B["/harness-plan"]
    B --> C["Plans.md"]
    C --> D["/harness-work"]
    D --> E["Code + Self-Review"]
    E --> F["/harness-review"]
    F --> G["Ship It"]

Requirements

Claude Code v2.1+ (Install Guide)
Node.js 18+ (for TypeScript core engine & safety hooks)

Install in 30 Seconds

# Start Claude Code in your project
claude

# Add the marketplace & install
/plugin marketplace add Chachamaru127/claude-code-harness
/plugin install claude-code-harness@claude-code-harness-marketplace

# Initialize your project
/harness-setup

That's it. Start with /harness-plan.

🪄 TL;DR: Work All

Don't want to read all this? Just type:

/harness-work all

One command. Harness does the rest. Plan → Parallel Implementation → Review → Commit.

Before	After
`/harness-plan` → `/harness-work` → `/harness-review` → `git commit`	`/harness-work all`
4 commands	1

⚠️ Experimental: Once you approve the plan, Claude runs to completion. Quality gate blocks commit if issues found.

The Core Loop

1. Plan

/harness-plan

"I want a login form with email validation"

Harness creates Plans.md with clear acceptance criteria.

2. Work

/harness-work              # Auto-detect parallelism
/harness-work --parallel 5 # 5 workers simultaneously

Each worker implements, self-reviews, and reports.

3. Review

/harness-review

Perspective	Focus
Security	Vulnerabilities, injection, auth
Performance	Bottlenecks, memory, scaling
Quality	Patterns, naming, maintainability
Accessibility	WCAG compliance, screen readers

Safety First

Harness v3 protects your codebase with a TypeScript guardrail engine (core/) — 9 declarative rules (R01–R09), compiled and type-checked:

Rule	Protected	Action
R01	`sudo` commands	Deny
R02	`.git/`, `.env`, secrets	Deny write
R03	`rm -rf /`, destructive paths	Deny
R04	`git push --force`	Deny
R05–R09	Mode-specific guards	Context-aware
Post	`it.skip`, assertion tampering	Warning
Perm	`git status`, `npm test`	Auto-allow

5 Verb Skills, Zero Config

v3 unifies 42 skills into 5 verb skills. Auto-load by context. Slash commands or natural language.

<table> <tr> <td align="center" width="20%"><h3>/plan</h3>Ideas → Plans.md</td> <td align="center" width="20%"><h3>/work</h3>Parallel implementation</td> <td align="center" width="20%"><h3>/review</h3>4-angle code review</td> <td align="center" width="20%"><h3>/release</h3>Tag + GitHub Release</td> <td align="center" width="20%"><h3>/setup</h3>Project init & config</td> </tr> </table> <p align="center"> <img src="assets/readme-visuals-en/skills-ecosystem.svg" alt="Skills ecosystem" width="640"> </p>

Key Commands

Command	What It Does	Legacy Redirect
`/harness-plan`	Ideas → `Plans.md`	`/plan-with-agent`, `/planning`
`/harness-work`	Parallel implementation	`/work`, `/breezing`, `/impl`
`/harness-work all`	Plan → Implement → Review → Commit	`/work all`
`/harness-review`	4-perspective code review	`/harness-review`, `/verify`
`/harness-release`	CHANGELOG, tag, GitHub Release	`/release-har`, `/handoff`
`/harness-setup`	Initialize project	`/harness-init`, `/setup`
`/memory`	Manage SSOT files	—

Who Is This For?

You Are	Harness Helps You
Developer	Ship faster with built-in QA
Freelancer	Deliver review reports to clients
Indie Hacker	Move fast without breaking things
VibeCoder	Build apps with natural language
Team Lead	Enforce standards across projects

Architecture

claude-code-harness/
├── core/           # TypeScript guardrail engine (strict ESM, NodeNext)
│   └── src/        #   guardrails/ state/ engine/
├── skills-v3/      # 5 verb skills (plan/execute/review/release/setup)
├── agents-v3/      # 3 agents (worker/reviewer/scaffolder)
├── hooks/          # Thin shims → core/ engine
├── skills/         # 41 legacy skills (retained for compatibility)
├── agents/         # 11 legacy agents (retained for compatibility)
├── scripts/        # v2 hook scripts (coexist with v3 core)
└── templates/      # Generation templates

Advanced Features

<details> <summary><strong>Breezing (Agent Teams)</strong></summary>

Run entire task lists with autonomous agent teams:

/harness-work breezing all                    # Plan review + parallel implementation
/harness-work breezing --no-discuss all       # Skip plan review, go straight to coding
/harness-work breezing --codex all            # Delegate to Codex engine

Phase 0 (Planning Discussion) runs by default—Planner analyzes task quality, Critic challenges the plan, then you approve before coding starts.

Feature	Description
Planning Discussion	Planner + Critic review your plan (default-on)
Task Validation (V1–V5)	Scope, ambiguity, overlap, dependency, TDD checks
Progressive Batching	8+ tasks auto-split into manageable batches
Hook-driven Signals	Auto-triggers for partial review and next batch

Cost: ~5.5x tokens (default) vs ~4x (with --no-discuss). The plan review pays for itself by reducing rework.

</details> <details> <summary><strong>Codex Engine</strong></summary>

Delegate implementation tasks to OpenAI Codex in parallel:

/harness-work --codex implement these 5 API endpoints

Codex implements → Self-reviews → Reports back. Works alongside Claude Code workers.

Setup required: Install Codex CLI and configure API key.

</details> <details> <summary><strong>Codex CLI Setup</strong></summary>

Use Harness with Codex CLI — no Claude Code required.

Prerequisites: Codex CLI (npm i -g @openai/codex), OpenAI API key (OPENAI_API_KEY), Git.

# 1. Clone the Harness repository
git clone https://github.com/Chachamaru127/claude-code-harness.git
cd claude-code-harness

# 2. Install skills/rules to user scope (~/.codex)
./scripts/setup-codex.sh --user

# 3. Go to your project and start working
cd /path/to/your-project
codex

Once inside Codex, use $harness-plan, $harness-work, $breezing, and $harness-review.

Flag	Description
`--user`	Install to `~/.codex` (shared across projects, default)
`--project`	Install to `.codex/` in current directory

Claude Code users can run /setup codex inside a session instead.

</details> <details> <summary><strong>2-Agent Mode (with Cursor)</strong></summary>

Use Cursor as PM, Claude Code as implementer.

/harness-release handoff  # Report to Cursor PM

Plans.md syncs between both.

</details> <details> <summary><strong>Codex Review Integration</strong></summary>

Add OpenAI Codex for second opinions:

/harness-review --codex  # 4 perspectives + Codex CLI

Codex selects 4 relevant experts from 16 specialist types via codex exec.

</details> <details> <summary><strong>Slide Generation</strong></summary>

Generate one-page project intro slides:

/generate-slide

3 visual patterns (Minimalist / Infographic / Hero)
2 candidates per pattern with quality scoring
Best 3 slides exported to out/slides/selected/

Dependencies: GOOGLE_AI_API_KEY and Google AI Studio access.

</details> <details> <summary><strong>Video Generation</strong></summary>

Generate product videos with JSON Schema-driven pipeline:

/generate-video

JSON Schema as SSOT (Single Source of Truth)
3-layer validation: scene → scenario → E2E
Remotion-based rendering with deterministic output

Dependencies: Requires Remotion project setup and ffmpeg.

</details> <details> <summary><strong>Agent Trace</strong></summary>

Automatically tracks AI-generated code edits:

.claude/state/agent-trace.jsonl

Records every Edit/Write operation
Shows project name, current task, recent edits at session end
Enables /sync-status to compare Plans.md with actual changes

No setup required—enabled by default.

</details>

Troubleshooting

Issue	Solution
Command not found	Run `/harness-setup` first
Plugin not loading	Clear cache: `rm -rf ~/.claude/plugins/cache/claude-code-harness-marketplace/` and restart
Hooks not working	Ensure Node.js 18+ is installed

For more help, open an issue.

Uninstall

/plugin uninstall claude-code-harness

Project files (Plans.md, SSOT files) remain unchanged.

Claude Code 2.1.69+ Features

Harness leverages the latest Claude Code features out of the box.

Feature	Skill	Purpose
Agent Memory	harness-work, harness-review	Persistent learning across sessions
TeammateIdle/TaskCompleted Hook	breezing	Automated team monitoring
Worktree isolation	breezing	Safe parallel writes to the same file
HTTP hooks	hooks	JSON POST to Slack, dashboards, metrics
Effort levels + ultrathink	harness-work	Auto-injects ultrathink for complex tasks
Agent hooks	hooks	LLM-powered code quality guards (secrets, TODO stubs, security)
`${CLAUDE_SKILL_DIR}` variable	all harness-* skills	Stable references to skill-local docs
InstructionsLoaded hook	hooks	Pre-session instruction-load tracking
`agent_id` / `agent_type` fields	hooks, breezing	Robust teammate identity and role guard
`{"continue": false}` teammate response	breezing	Auto-stop when all assigned tasks are complete
`/reload-plugins`	all harness-* skills	Apply skill/hook edits immediately
`includeGitInstructions: false`	harness-work, breezing	Lower token cost on git-light tasks
WorktreeCreate/Remove hook	breezing	Worktree lifecycle auto-setup and cleanup

Full list: docs/CLAUDE-feature-table.md

Documentation

Resource	Description
Changelog	Version history
Claude Code Compatibility	Requirements
Cursor Integration	2-Agent setup

Contributing

Issues and PRs welcome. See CONTRIBUTING.md.

Acknowledgments

AI Masao — Hierarchical skill design
Beagle — Test tampering prevention patterns

License

MIT License — Free to use, modify, commercialize.

English | 日本語

Name	ultrawork
Description	Autonomously iterates until specified Plans.md range is complete - long-running /work with self-learning. Use when user mentions '/ultrawork', complete until done, finish all tasks, or autonomous execution. Do NOT load for: single tasks, reviews, or setup.

ultrawork

SKILL.md