Agent-Harness

# Agent-Harness

[![CI](https://github.com/zuoyui/Agent-Harness/actions/workflows/ci.yml/badge.svg)](https://github.com/zuoyui/Agent-Harness/actions/workflows/ci.yml)
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
[![Use this template](https://img.shields.io/badge/Use%20this%20template-2ea043?logo=github)](https://github.com/zuoyui/Agent-Harness/generate)

**Turn any repository into an agent-readable, governed, feedback-looped workspace.**

*An agent-agnostic harness engineering framework for AI coding agents.*

> Most AI coding agents fail not because the model is weak, but because the repository is unreadable, under-specified, and missing executable constraints. Agent-Harness fixes this.

## Framework Overview

```mermaid
flowchart TB
    A["Agent-Harness"] --> B["Core Layer<br/>AGENTS, architecture, docs, governance"]
    A --> C["Adapter Layer<br/>Windsurf, Codex, Claude Code, Cursor"]
    A --> D["Example Layer<br/>Next.js, Python service, Monorepo"]
```

## Quickstart

### Option 1 — Use as template

[**→ Use this repository as a template**](https://github.com/zuoyui/Agent-Harness/generate)

### Option 2 — Add to an existing project

```bash
cp -r core/ your-project/
cp -r adapters/codex/.codex your-project/
cp -r core/governance/* your-project/.github/
```

## Supported Agents

| Agent | Status |
|-------|--------|
| [Windsurf](adapters/windsurf/) | ✅ Full implementation |
| [Codex](adapters/codex/) | ✅ Full implementation |
| [Claude Code](adapters/claude-code/) | ✅ Full implementation |
| [Cursor](adapters/cursor/) | 📖 Guide available |
| [Other agents](adapters/generic/) | 📖 Generic guide |

## Examples

| Example | Scenario |
|---------|----------|
| [Next.js App](examples/nextjs-app/) | Frontend/full-stack with Tailwind and TypeScript |
| [Python Service](examples/python-service/) | Backend service with security constraints |
| [Monorepo](examples/monorepo/) | Multi-package project with neste