By evanfang0054
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Use when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always
Use when you have a spec or requirements for a multi-step task, before touching code
Use during retrospective or when trying to understand recurring failure patterns across sessions, based on historical learnings data in .agent-harness/learnings.jsonl.
Use when starting any conversation - establishes how to find and use skills, requiring Skill tool invocation before ANY response including clarifying questions
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior. Explores user intent, requirements and design before implementation.
Executes bash commands
Hook triggers when Bash tool is used
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Agent Harness 是一套完整的软件开发工作流,专为 AI 编程助手设计。它基于一系列可组合的"技能(skills)"构建,并通过初始指令确保你的 AI 助手能够正确使用它们。
当你启动 AI 编程助手时,它不会立即开始写代码。相反,它会退一步,先问你真正想要实现什么。
在通过对话梳理出需求后,它会将设计方案分成易于阅读和理解的小块展示给你。
在你确认设计方案后,AI 助手会制定一个实现计划,这个计划清晰到即使是一个热情但缺乏经验、没有项目背景、不爱写测试的初级工程师也能遵循。它强调真正的红绿测试驱动开发(TDD)、YAGNI(你不会需要它)和 DRY(不要重复自己)原则。
接下来,当你说"开始"后,它会启动子代理驱动开发流程,让多个 AI 代理协作完成每个工程任务,检查和审查它们的工作,然后继续推进。Claude 通常可以自主工作几个小时而不偏离你们共同制定的计划。
这只是系统的核心部分,还有更多功能。由于技能会自动触发,你不需要做任何特别的事情。你的 AI 编程助手就拥有了 Agent Harness。
注意: 不同平台的安装方式不同。Claude Code 和 Cursor 有内置的插件市场,Codex 和 OpenCode 需要手动设置。
Agent Harness 可通过 Claude 官方插件市场 获取
从 Claude 市场安装插件:
/plugin install agent-harness@claude-plugins-official
在 Claude Code 中,先注册市场:
/plugin marketplace add evanfang0054/agent-harness-marketplace
然后从该市场安装插件:
/plugin install agent-harness@agent-harness-marketplace
在 Cursor Agent 聊天中,从市场安装:
/add-plugin agent-harness
或在插件市场中搜索 "agent-harness"。
在 Codex App 侧边栏点击 Plugins,在 Coding 区找到 Agent Harness,点击 + 安装。
打开插件搜索界面 /plugins,搜索 agent-harness,选择 Install Plugin。
详细文档: docs/README.codex.md
从 GitHub 仓库安装:
pi install git:github.com/evanfang0054/agent-harness
本地开发模式:
pi -e /path/to/agent-harness
详细文档: docs/README.pi.md
告诉 OpenCode:
Fetch and follow instructions from https://raw.githubusercontent.com/evanfang0054/agent-harness/refs/heads/main/.opencode/INSTALL.md
详细文档: docs/README.opencode.md
copilot plugin marketplace add evanfang0054/agent-harness-marketplace
copilot plugin install agent-harness@agent-harness-marketplace
gemini extensions install https://github.com/evanfang0054/agent-harness
更新:
gemini extensions update agent-harness
在你选择的平台上启动新会话,请求一些应该触发技能的操作(例如,"帮我规划这个功能"或"让我们调试这个问题")。AI 助手应该会自动调用相关的 agent-harness 技能。
Agent Harness 采用分层架构:决策层确保"做对的事",执行层确保"把事做对",质量层确保"做得好"。
┌─────────────────────────────────────────────────────────────────────────────┐
│ 决策层(Decision Layer) │
│ "该不该做?怎么定方向?" │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌─────────────┐ ┌─────────────────┐ ┌─────────────────┐ │
│ │office-hours │ ───► │ plan-ceo-review │ ───► │ plan-eng-review │ │
│ │ "值得做吗?" │ │ "10星产品?" │ │ "架构可行?" │ │
│ └─────────────┘ └─────────────────┘ └─────────────────┘ │
│ │ │ │
│ │ 🟢 值得做 │ ✅ 架构锁定 │
│ ▼ ▼ │
└─────────────────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────────────────┐
│ 执行层(Execution Layer) │
│ "怎么设计?怎么实现?" │
├─────────────────────────────────────────────────────────────────────────────┤
│ │
│ ┌─────────────┐ ┌────────────────┐ ┌─────────────┐ ┌───────────────┐ │
│ │brainstorming│─►│ gate-driven- │─►│writing-plans│─►│subagent-dev / │ │
│ │ "怎么设计?" │ │ test-design * │ │ "拆分任务" │ │ exec-plans │ │
│ └─────────────┘ └────────────────┘ └─────────────┘ └───────────────┘ │
│ * 可选:递归生成 │
│ 测试金字塔 │
│ │ │
│ ┌─────────────────────────────────────────────────────┼───────────────┐ │
│ │ 实现循环 ▼ │ │
│ │ ┌─────┐ ┌───────────┐ ┌─────────────┐ ┌──────┐ ┌─────────┐ │ │
│ │ │ TDD │─►│comp-sensor│─►│code-review │─►│verify│─►│finishing│ │ │
│ │ └─────┘ └───────────┘ └─────────────┘ └──────┘ └─────────┘ │ │
│ └─────────────────────────────────────────────────────────────────────┘ │
│ │
│ ┌─────────────────────────────────────────────────────────────────────┐ │
│ │ harness-init (初始化) · harness-design (原型) · harness-optimizer │ │
npx claudepluginhub evanfang0054/agent-harness --plugin agent-harnessHarness-native ECC plugin for engineering teams - 67 agents, 277 skills, 93 legacy command shims, reusable hooks, rules, MCP conventions, and operator workflows for Claude Code plus adjacent agent harnesses
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Supergraph enforces a complete, evidence-based coding pipeline — scan → plan → TDD → fix → verify → review — grounded in real codebase analysis at every step. It combines AST dependency graphs, LSP-level code intelligence, and a structured skill chain so Claude never guesses about impact before making a change.
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.