A
AhmedAbdelhady24/engineering-workflow
Universal multi-stage software engineering workflow skill for Google Antigravity, DeepSeek Harness, OpenAI…
在 GitHub 查看 ↗链接已于 2026-11-12 核查 · HTTP 200 · 打开的是厂商自己的官网
安装与配置
客户端配置（取自该仓库的安装命令）
{
"mcpServers": {
"ahmedabdelhady24-engineering-workflow": {
"command": "npx",
"args": ["skills"]
}
}
}解析自我们于 2026-10-28 抓取的该仓库 README；我们未执行过。
command/args 来自仓库写明的安装命令；不确定时请以该项目 README 为准。
VS Code（.vscode/mcp.json —— 键名是 servers）
{
"servers": {
"ahmedabdelhady24-engineering-workflow": {
"command": "npx",
"args": ["skills"]
}
}
}VS Code 顶层键是 <code>servers</code>，其余与上面那份相同。
Claude Code（命令行）
claude mcp add --transport stdio ahmedabdelhady24-engineering-workflow -- npx skills
Claude Code 不收 JSON 文件，用这条命令直接加。
配置项的键名以你的客户端为准：客户端与 CLI
- ★ 数取值
- 0 ★ · AhmedAbdelhady24/engineering-workflow · 取自 GitHub 搜索 API · （2026-11-13 20:00:50 核对）
- 来源明细
- 从该仓库的 GitHub 页面直接读取 · 条目类型：开源项目
- 收录与活跃
- 收录于 2026-10-28 · 最近见到 2026-11-14 · 共见到 19 次 · 被 1 个来源列出
- 评分构成
- 综合分 103 · 1 个来源 · repo
- 安装
npx skills
常见问题（答案都来自我们的记录）
怎么安装？
该仓库给出的安装命令是
npx skills。我们的记录里它是仓库自己提供的；不保证在你的客户端上直接可用。它适合谁 / 用来做什么？
我们把它归在这些标签下: Agent 技能。标签来自条目自身，不是我们强加的分类。
谁维护它、还在更新吗？
仓库
AhmedAbdelhady24/engineering-workflow；我们收录于 2026-10-28；最近见到 2026-11-14。这里没有"官方认证"或"质量评级"——我们只记录来源与时间。# Universal Engineering Workflow
以下摘录自该仓库的 README（片段） · 在 GitHub 读完整 README →
The compaction-resilient orchestrator for real software engineering — not vibe coding.
Developing production applications with coding agents is hard. Monolithic frameworks try to help by owning the process, locking your terminal into rigid state machines and stripping away human control. At the other extreme, unguided agents fly without a compass, hallucinate passing tests, and suffer from total amnesia the moment context compacts.
Universal Engineering Workflow provides the missing discipline: four lean operations (setup, execute, review, ux-audit), an ironclad Skill Execution Contract, a machine-readable Skill Registry, a compaction-resilient recovery ledger, and end-to-end runtime verification across Claude Code, OpenAI Codex, Google Antigravity, and DeepSeek Harness.
## What It Does vs. What It Does NOT Do
| What This Workflow DOES | What This Workflow Does NOT Do |
| Enforces four disciplined operations: setup, execute, review, and ux-audit. | Does NOT hijack your terminal or run proprietary background daemon processes. |
| Enforces the Skill Execution Contract: Verifies installed skills, invokes them explicitly, and requires hard command evidence. | Does NOT vendor third-party skills: References canonical upstreams (DietrichGebert/ponytail, mattpocock/skills, obra/superpowers, addyosmani/agent-skills). |
| Maintains native test discipline: Applies red-green cycles natively even when external TDD skills are absent. | Does NOT claim unverified execution: Never claims TDD or Ponytail was invoked without real execution evidence. |
| Provides compaction-resilient recovery: Persists checkpoints in TASKLOG.md and TASKRECOVERYLEDGER.md with explicit git and test reconciliation protocols. | Does NOT lock you into proprietary runtimes: Pure markdown, JSON manifests, and standard shell scripts. |
| Conducts two-axis code reviews: Inspects diffs for both Engineering Standards and Spec fulfillment. | Does NOT mutate code during reviews or audits: Review and UX-audit roles are strictly read-only. |
| Performs 8-stage UX audits: Walks real user journeys, audits 10 Nielsen heuristics, and inspects accessibility. | Does NOT equate automated passes with full WCAG conformance: Distinguishes automated checks from true accessibility. |
## Prerequisites
Git: v2.20 or later with standard porcelain support.
Node.js: v18.0.0 or later (Node 20+ recommended for test suite).
Supported Agent Environments:
Claude Code (CLI or plugin marketplace)
OpenAI Codex (CLI or IDE extension)
Google Antigravity / Gemini CLI (agy)
DeepSeek Harness (DSH)
Any agent supporting the open Agent Skills specification (Cursor, Windsurf, OpenCode)
## 30-Second Setup
Pick your agent environment below:
### 1. Install
Claude Code (Plugin Marketplace)
bash
claude plugin marketplace add AhmedAbdelhady24/engineering-workflow
claude plugin install engineering-workflow@AhmedAbdelhady24
Follows Claude Code plugin marketplace update policies.
OpenAI Codex (Plugin Marketplace)
bash
codex plugin marketplace add AhmedAbdelhady24/engineering-workflow
codex plugin add engineering-workflow@AhmedAbdelhady24
Follows Codex plugin marketplace update policies.
Google Antigravity / Gemini CLI (Native Skills)
Developing production applications with coding agents is hard. Monolithic frameworks try to help by owning the process, locking your terminal into rigid state machines and stripping away human control. At the other extreme, unguided agents fly without a compass, hallucinate passing tests, and suffer from total amnesia the moment context compacts.
Universal Engineering Workflow provides the missing discipline: four lean operations (setup, execute, review, ux-audit), an ironclad Skill Execution Contract, a machine-readable Skill Registry, a compaction-resilient recovery ledger, and end-to-end runtime verification across Claude Code, OpenAI Codex, Google Antigravity, and DeepSeek Harness.
## What It Does vs. What It Does NOT Do
| What This Workflow DOES | What This Workflow Does NOT Do |
| Enforces four disciplined operations: setup, execute, review, and ux-audit. | Does NOT hijack your terminal or run proprietary background daemon processes. |
| Enforces the Skill Execution Contract: Verifies installed skills, invokes them explicitly, and requires hard command evidence. | Does NOT vendor third-party skills: References canonical upstreams (DietrichGebert/ponytail, mattpocock/skills, obra/superpowers, addyosmani/agent-skills). |
| Maintains native test discipline: Applies red-green cycles natively even when external TDD skills are absent. | Does NOT claim unverified execution: Never claims TDD or Ponytail was invoked without real execution evidence. |
| Provides compaction-resilient recovery: Persists checkpoints in TASKLOG.md and TASKRECOVERYLEDGER.md with explicit git and test reconciliation protocols. | Does NOT lock you into proprietary runtimes: Pure markdown, JSON manifests, and standard shell scripts. |
| Conducts two-axis code reviews: Inspects diffs for both Engineering Standards and Spec fulfillment. | Does NOT mutate code during reviews or audits: Review and UX-audit roles are strictly read-only. |
| Performs 8-stage UX audits: Walks real user journeys, audits 10 Nielsen heuristics, and inspects accessibility. | Does NOT equate automated passes with full WCAG conformance: Distinguishes automated checks from true accessibility. |
## Prerequisites
Git: v2.20 or later with standard porcelain support.
Node.js: v18.0.0 or later (Node 20+ recommended for test suite).
Supported Agent Environments:
Claude Code (CLI or plugin marketplace)
OpenAI Codex (CLI or IDE extension)
Google Antigravity / Gemini CLI (agy)
DeepSeek Harness (DSH)
Any agent supporting the open Agent Skills specification (Cursor, Windsurf, OpenCode)
## 30-Second Setup
Pick your agent environment below:
### 1. Install
Claude Code (Plugin Marketplace)
bash
claude plugin marketplace add AhmedAbdelhady24/engineering-workflow
claude plugin install engineering-workflow@AhmedAbdelhady24
Follows Claude Code plugin marketplace update policies.
OpenAI Codex (Plugin Marketplace)
bash
codex plugin marketplace add AhmedAbdelhady24/engineering-workflow
codex plugin add engineering-workflow@AhmedAbdelhady24
Follows Codex plugin marketplace update policies.
Google Antigravity / Gemini CLI (Native Skills)
收录徽章（可选）
想把收录标记放进 README 就复制这行，可选，不影响收录。
[](https://anclum.com/zh/s/ahmedabdelhady24-engineering-workflow/)
- 来源
- github · https://github.com/AhmedAbdelhady24/engineering-workflow
- 按领域浏览
- Agent 技能
- 收录于
- 今天
同类推荐
所属领域与标签
每个标签都能点进去看同类的全部条目。数字是当前库里的真实条数。