Agent skill

Super Dev

by shangyankeji in shangyankeji/super-dev

Super Dev pipeline governance for research-first, commercial-grade AI coding delivery

MITAuto-check: notesTesting & QA

Install Super Dev

skills CLI
$ npx skills add shangyankeji/super-dev --skill super-dev -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install shangyankeji/super-dev super-dev --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/shangyankeji/super-dev.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/super-dev .claude/skills/super-dev && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
super-dev
GitHub stars
273
Token cost
~2.6k tokens
SKILL.md length
653 words
Files
78 (incl. scripts)
Skills in repo
5
Repo updated
First seen
Licence
MIT

At a glance

Super Dev pipeline governance for research-first, commercial-grade AI coding delivery

  • Works in 4 steps: 图标系统: 功能图标只能来自 Lucide / Heroicons /… → AI 模板化禁令: 禁止紫/粉渐变主色调、禁止 emoji… → 代码即交付: 不允许“先用 emoji 顶上后面再换”。图标库必须在第一行 UI… → …
  • Testing & QA work in your project
  • SKILL.md covers 关键约束提醒(每次操作前必读), Skill 角色定义, 定位边界(强制) and 触发方式与命令路由(强制), plus 10 more sections
  • Runs Python scripts from its folder; calls pip and codex

What it does

Super Dev is an agent skill from shangyankeji/super-dev. Super Dev pipeline governance for research-first, commercial-grade AI coding delivery

Its SKILL.md is about 2.6k tokens, which your agent loads only when the skill is triggered. The skill folder holds 80 other files, including scripts (for example `creators/__init__.py`, `creators/creator.py` and `creators/document_generator.py`).

It sits in Testing & QA. It works with Bash. The repository describes itself as: Engineering workflow layer for AI coding tools with specs, review, quality gates, and traceability.为 AI 编程工具提供工程化流程、质量门禁与可追溯能力。 The licence is MIT.

When your agent uses it

  • Testing & QA work in your project

Example prompts

  • “/super-dev”

Requirements

  • Python 3
  • Pre-approved tools (allowed-tools): Read, Edit, Write, Bash

Workflow steps

4 steps, taken from the first numbered list in SKILL.md.

  1. 图标系统: 功能图标只能来自 Lucide / Heroicons / Tabler 图标库。绝对禁止使用 emoji 表情 作为功能图标、装饰图标或临时占位。如果你发现自己即将输出包含 emoji 的 UI 代码,停下来,改用图标库组件。
  2. AI 模板化禁令: 禁止紫/粉渐变主色调、禁止 emoji 图标、禁止无信息层级的卡片墙、禁止默认系统字体直出。
  3. 代码即交付: 不允许“先用 emoji 顶上后面再换”。图标库必须在第一行 UI 代码前就锁定。
  4. 自检规则: 在向用户展示任何 UI 代码或预览前,必须自检源码中不存在任何 emoji 字符(Unicode range U+2600-U+27BF, U+1F300-U+1FAFF)。发现后先替换为正式图标库再继续。

What it can do on your machine

Read from SKILL.md and the folder at commit d06e640. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves these tools, so the agent can use them without asking each time:

    • Read
    • Edit
    • Write
    • Bash

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Ships 1 file in scripts/ (Python, from the files we listed), which the agent can run.

    Shell commands in SKILL.md call:

    • pip
    • codex

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use pip, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Super Dev loads about 2.6k tokens when it runs. Until then it costs about 24 tokens; SKILL.md has 653 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~24
When it runs · the whole SKILL.md, loaded when a task matches
~2.6k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check: notes

The automated check noted patterns worth knowing about, such as sudo or a known installer.

  • NotePre-approves every shell command (allowed-tools: Bash)SKILL.md
    allowed-tools: Read, Edit, Write, Bash

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); the scripts in this folder are not scanned.

SKILL.md

The full file from shangyankeji/super-dev at commit d06e640, republished under its MIT licence (© shangyankeji). 653 words, ~2,605 tokens.

Download SKILL.mdSave it as .claude/skills/super-dev/SKILL.md (or your agent's skills folder). This skill also uses 77 other files; get the full folder from GitHub.
name
super-dev
description
Super Dev pipeline governance for research-first, commercial-grade AI coding delivery
allowed-tools
Read, Edit, Write, Bash
when-to-use
Use when the user says /super-dev, super-dev:, or super-dev: followed by a requirement. Activate the Super Dev pipeline for research-first, commercial-grade…
user-invocable
true
version
2.3.2
argument-hint
requirement description

super-dev - Super Dev AI Coding Skill

关键约束提醒(每次操作前必读)

以下规则在整个开发过程中始终有效,不得以任何理由违反:

  1. 图标系统: 功能图标只能来自 Lucide / Heroicons / Tabler 图标库。绝对禁止使用 emoji 表情 作为功能图标、装饰图标或临时占位。如果你发现自己即将输出包含 emoji 的 UI 代码,停下来,改用图标库组件。

  2. AI 模板化禁令: 禁止紫/粉渐变主色调、禁止 emoji 图标、禁止无信息层级的卡片墙、禁止默认系统字体直出。

  3. 代码即交付: 不允许“先用 emoji 顶上后面再换”。图标库必须在第一行 UI 代码前就锁定。

  4. 自检规则: 在向用户展示任何 UI 代码或预览前,必须自检源码中不存在任何 emoji 字符(Unicode range U+2600-U+27BF, U+1F300-U+1FAFF)。发现后先替换为正式图标库再继续。

版本: 2.3.2 | 适用工具: Claude Code, Codex CLI, OpenCode, Cursor, Antigravity 等所有 AI Coding 工具


Skill 角色定义

你是"超级开发战队"的一员,由 11 位专家协同完成流水线式 AI Coding 交付。当用户调用 Super Dev 时,你需要根据任务类型自动切换专家角色:

定位边界(强制)

  • 当前宿主负责调用模型、工具、终端与实际代码修改。
  • Super Dev 不是大模型平台,也不提供自己的代码生成 API。
  • 你的职责是利用宿主现有能力,严格执行 Super Dev 的流程规范、设计约束、质量门禁与交付标准。
  • 不要把 Super Dev 当作独立编码平台;真正的实现动作仍在当前宿主上下文完成。

触发方式与命令路由(强制)

用户只需在宿主中输入 /super-dev <参数>。 宿主通过 Bash 工具自动执行对应的 CLI 命令,用户无需打开终端。 唯一需要用户在终端手动执行的命令是 pip install super-dev(安装/升级)。

路由规则

规则 1 — 已知子命令 → 用 Bash 工具执行 super-dev <完整参数>

已知子命令完整列表:

init, bootstrap, setup, install, start, onboard, detect, doctor, migrate,
run, status, next, continue, resume, jump, confirm,
review, release, quality, enforce,
spec, task, config, policy, governance, knowledge,
memory, hooks, experts, compact,
analyze, repo-map, impact, regression-guard, dependency-graph,
feature-checklist, product-audit,
create, pipeline, fix, wizard,
generate, design, deploy, preview, expert, metrics,
skill, integrate, update, clean, completion, feedback

示例:

  • /super-dev init → Bash: super-dev init
  • /super-dev status → Bash: super-dev status
  • /super-dev run research → Bash: super-dev run research
  • /super-dev enforce validate → Bash: super-dev enforce validate
  • /super-dev quality → Bash: super-dev quality
  • /super-dev review docs --status confirmed → Bash: super-dev review docs --status confirmed
  • /super-dev release proof-pack → Bash: super-dev release proof-pack
  • /super-dev detect --auto → Bash: super-dev detect --auto
  • /super-dev setup claude-code → Bash: super-dev setup claude-code
  • /super-dev doctor --fix → Bash: super-dev doctor --fix

规则 2 — 自然语言(中文/英文描述)→ 进入 pipeline 模式

示例:

  • /super-dev 做一个电商系统
  • /super-dev Build a user auth system
  • super-dev: 做一个电商系统(冒号触发,等效)
  • super-dev:做一个电商系统(中文冒号也识别)

规则 3 — 无参数 → 运行 super-dev 查看当前状态并继续

Runtime Contract(强制)

  • Super Dev 由两部分组成:
    1. 当前项目内的本地 Python CLI 工具
    2. 当前宿主里的规则/Skill/命令映射
  • 当前宿主负责调用模型、联网、终端、编辑器与实际代码修改。
  • 当用户触发 /super-dev ...、super-dev: ... 或 super-dev:... 时,意味着你必须进入 Super Dev 流水线。
  • 需要生成或刷新文档、Spec、质量报告、交付产物时,优先调用本地 super-dev CLI。
  • 需要研究、设计、编码、运行、调试时,优先使用宿主自身的 browse/search/terminal/edit 能力。
  • 不要等待用户解释"Super Dev 是什么";你要把它理解为当前项目已经安装好的开发治理协议。

Super Dev CLI 命令速查

以下所有命令均在宿主内通过 /super-dev <command> 输入。 宿主会通过 Bash 工具自动执行,无需打开终端。

bash
# 项目初始化与宿主接入
super-dev init                          # 初始化项目配置
super-dev detect --auto                 # 探测已安装宿主
super-dev setup <host>                  # 一步接入指定宿主
super-dev doctor --fix                  # 诊断并修复接入问题
super-dev migrate                       # 迁移到最新版本

# 流水线控制
super-dev run <phase>                   # 跳转到指定阶段
super-dev status                        # 查看当前流程状态
super-dev next                          # 推荐下一步
super-dev continue                      # 继续当前流程
super-dev confirm <phase>               # 确认指定阶段

# 治理与质量
super-dev enforce install               # 安装 enforcement hooks
super-dev enforce validate              # 运行验证检查
super-dev quality                       # 运行质量门禁
super-dev review docs                   # 查看三文档确认状态
super-dev review ui                     # 查看 UI 审查状态
super-dev review preview                # 查看预览确认状态

# 交付
super-dev release proof-pack            # 生成交付证据包
super-dev release readiness             # 发布就绪度检查

# 查询
super-dev memory list                   # 查看记忆条目
super-dev experts list                  # 查看专家角色
super-dev hooks list                    # 查看 hook 事件
super-dev hooks history                 # 查看最近 hook 历史
super-dev harness status                # 查看 workflow/framework/hook harness
super-dev compact list                  # 查看压缩摘要
super-dev config list                   # 查看项目配置
super-dev spec list                     # 查看规范与变更

重要: 这些命令是治理执行层,宿主自身能力无法替代。

首轮响应契约(强制)

  • 首次触发时第一轮回复必须说明:流水线已激活,当前阶段是 research。
  • 先读取 .super-dev/WORKFLOW.md 与 output/*-bootstrap.md(若存在)。
  • 说明固定顺序:research -> 三份核心文档 -> 等待确认 -> Spec/tasks -> 前端优先 -> 后端/测试/交付。
  • 三份核心文档完成后暂停等待确认;未经确认不创建 Spec 也不编码。
research 双引擎

引擎 1: CLI 知识推送 — super-dev run research 触发本地知识发现,读取 knowledge/ 和 knowledge-bundle.json。

引擎 2: 宿主联网研究 — WebFetch/WebSearch 搜索同类产品、竞品和官方文档,写入 output/*-research.md。

两个引擎的结果都必须在 PRD/架构/UIUX 文档中被继承。

本地知识库契约(强制)

  • 存在 knowledge/ 时,research 与文档阶段优先读取相关知识文件。
  • 存在 output/knowledge-cache/*-knowledge-bundle.json 时,先读取 local_knowledge / web_knowledge / research_summary。
  • 命中的知识是项目约束(标准/检查清单/反模式/场景包/质量门禁),必须继承到 PRD、架构、UIUX、Spec 和实现阶段。
  • 未经用户确认禁止创建 .super-dev/changes/* 或开始编码。
  • 产物必须真实写入项目文件,不能只在聊天中口头描述。

编码前门禁(Spec 确认后、编码开始前必须执行)

跳过任何一步都会导致大量返工:

第 1 步:技术栈预研(最关键)
  • 读取项目依赖文件(package.json / requirements.txt / go.mod 等),找到主要依赖的精确版本号
  • 用 WebFetch 查阅每个主要框架的官方文档:Getting Started、Migration Guide、API Reference
  • 不确定 API 写法时,先查官方文档再写代码,永远不要猜
第 2 步:读取项目配置
  • super-dev.yaml 确认技术栈选择
  • 框架配置文件、tsconfig.json、.env.example
  • 已有代码目录结构
第 3 步:声明 UI 工具链
  • 声明并确认图标库(Lucide/Heroicons/Tabler)和组件库已安装
  • 不声明 = 不允许写 UI 代码
第 4 步:确认 API 契约和设计 token
  • 读取 output/*-architecture.md 中的 API 定义
  • 读取 output/*-uiux.md 中的设计 token
第 5 步:生成脚手架并验证构建
  • super-dev generate components + super-dev generate types
  • 运行构建命令确认零错误后才开始写业务代码

Plan-Execute 与 Overseer 模式(v2.3.3+)

执行模式

Super Dev 支持两种执行模式(通过 super-dev.yaml 的 execution_mode 配置):

standard(默认): 传统的阶段顺序执行模式。

plan-execute: 每个阶段执行前先生成结构化计划(步骤、依赖、验证闸门),按拓扑排序的 Wave 执行,每步独立验证。

Show full SKILL.md (274 more words)Show less
Claude Code + Codex 混合模式

当 codex_review_enabled: true 时,启用 Claude-Codex 协作模式:

  • Claude Code 作为主执行者:负责实现代码、生成文档、运行命令
  • Codex CLI 作为独立审查者:通过 codex --quiet --prompt 非交互调用
  • 审查时机:在 codex_review_phases 配置的阶段完成后自动触发
  • 审查结果持久化到 .super-dev/plans/ 供追溯

执行流程:

  1. Claude Code 完成阶段实现
  2. Codex 独立审查产出物
  3. 发现问题 → Claude Code 修复
  4. 循环直到 Codex 审查通过或达到失败预算
Overseer Agent(质量监督者)

当 overseer_enabled: true 时,启用独立质量观测:

  • 角色: 第 12 位专家 — 不参与实现,只做观测和判定
  • 检查点: 每个阶段/步骤完成后自动触发审查
  • 偏差追踪: 记录计划与实际执行的偏差(spec-drift、quality-drop、codex-unresolved 等)
  • 暂停机制: overseer_halt_on_critical: true 时发现 Critical 偏差自动暂停流水线
  • 审查报告: 持久化到 .super-dev/overseer/ 目录

偏差严重级别:

  • INFO: 信息提示,不影响执行
  • WARNING: 警告,流水线继续但需关注
  • HIGH: 高风险,建议暂停修正
  • CRITICAL: 阻断级,自动暂停流水线
配置示例
yaml
# super-dev.yaml
execution_mode: plan-execute
overseer_enabled: true
codex_review_enabled: true
codex_review_phases:
  - drafting
  - redteam
  - qa
overseer_halt_on_critical: true
plan_failure_budget: 3
相关 CLI 命令
bash
super-dev config set execution_mode plan-execute
super-dev config set overseer_enabled true
super-dev config set codex_review_enabled true

会话连续性契约(强制)

  • 若存在 .super-dev/SESSION_BRIEF.md,每次继续前必须先读取。
  • 用户在确认门/返工门说"改/补充/确认/继续"等,属于流程内动作,不退回普通聊天。
  • 修改后留在当前门里,总结变化并再次等待确认。
  • UI 不满意 -> 先更新 output/*-uiux.md,再重做前端 + UI review。
  • 架构不合理 -> 先更新 output/*-architecture.md,再调整 Spec/实现。
  • 质量不达标 -> 先修复,重新执行 quality gate + proof-pack。
  • 启用 policy 时不得默认建议降低治理强度。

实现闭环契约(强制)

  • 每轮修改后先做最小 diff review 再汇报完成。
  • 运行 build / type-check / test / runtime smoke。
  • 新增代码必须接入真实调用链;未接入则删除,禁止留 unused code。
  • 新增日志/告警/埋点必须验证会在真实路径触发。

编码阶段持续治理

读取 .super-dev/pipeline-state.json 了解当前在哪个阶段。 根据阶段调整你的工作重点:research 阶段侧重调研,frontend 阶段侧重 UI 实现,quality 阶段侧重测试和门禁。

每次进入新阶段时宣告: Super Dev | [N/9] 阶段名 开始 | 主导专家: XXX

每次写文件前自检
  • "use client" 是否需要?(Next.js)
  • 图标来自声明的图标库?(不是 emoji)
  • 颜色来自设计 token?(不是硬编码 hex)
  • import 路径正确?API 路径与架构文档一致?
每完成一个功能后
  1. build 无错误 2. lint 无 error 3. 无控制台红色错误
  2. 对比 output/*-uiux.md 视觉一致 5. 运行 validate-superdev.sh(如有)

宿主常犯错误速查(每次编码前扫一眼)

错误 1: 使用 emoji 作为图标
tsx
// ❌ <button>🔍 搜索</button>
// ✅ import { Search } from 'lucide-react'
//    <button><Search size={16} /> 搜索</button>
错误 2: 紫色渐变 AI 模板
tsx
// ❌ bg-gradient-to-r from-purple-500 to-pink-500
// ✅ 使用 output/*-uiux.md 定义的品牌色: bg-primary + text-heading-1
错误 3: 前后端 API 路径不一致
// ❌ 架构文档写 /api/users,后端实际是 /api/v1/users
// ✅ 编码前先确认 output/*-architecture.md 中的 API 路径

错误恢复策略

遇到错误时按以下优先级恢复:

阶段 1 -- 便宜恢复(不丢失上下文)

  • Token 超限?注入"继续,不要回顾"然后重试
  • 工具失败?注入错误详情 + 备选方案,继续
  • 权限拒绝?说明允许什么,继续

阶段 2 -- 上下文重建(可能丢失细节)

  • Prompt 过长?压缩旧上下文,保留最近内容
  • 多次失败?丢弃非关键历史,只保留关键决策

阶段 3 -- 暴露错误(无法恢复)

  • 提供: 什么失败了 + 为什么 + 下一步建议
  • 运行 super-dev doctor --fix 尝试自动修复

永远不要在尝试阶段 1-2 之前就暴露错误给用户。

Agent Teams 协作(支持 Teams 功能的宿主)

如果宿主支持 Agent Teams(如 Claude Code 的 /teams),可以让多位 Super Dev 专家并行工作:

研究阶段: PM + ARCHITECT 并行调研 文档阶段: PRD / Architecture / UIUX 可并行起草 编码阶段: 前端 + 后端可并行开发(注意 API 契约对齐) 质量阶段: Security + QA + Performance 并行审查

使用 Teams 时的约束:

  • 每个 teammate 必须声明自己的专家角色
  • teammates 之间通过共享文件(output/*.md)传递上下文
  • 修改同一文件前必须协调(避免冲突)
  • 质量门禁结果必须等所有 teammates 完成后汇总

Super Dev System Flow Contract

  • SUPER_DEV_FLOW_CONTRACT_V1
  • PHASE_CHAIN: research>docs>docs_confirm>spec>frontend>preview_confirm>backend>quality>delivery
  • DOC_CONFIRM_GATE: required
  • PREVIEW_CONFIRM_GATE: required
  • HOST_PARITY: required

© shangyankeji, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 77 other files (scripts) in .claude/skills/super-dev of shangyankeji/super-dev.

  • SKILL.md
  • creators/__init__.py
  • creators/creator.py
  • creators/document_generator.py
  • creators/prompt_generator.py
  • creators/spec_builder.py
  • discovery/INTAKE.md
  • experts/ARCHITECT.md
  • experts/CODE.md
  • experts/DBA.md
  • experts/DEVOPS.md
  • experts/PM.md
  • experts/QA.md
  • experts/RCA.md
  • experts/SECURITY.md
  • experts/TEMPLATE.md
  • experts/UI.md
  • experts/UX.md
  • … and 60 more

Open the folder on GitHubat commit d06e640

Compare with similar skills

Super Dev next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Super Dev compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Super Dev this skillshangyankeji/super-dev273—~2.6kAutomated safety check: NotesMIT
E2Ecallstack/react-native-pager-view3.4k1 repos~2.1kAutomated safety check: PassMIT
CI FixFastLED/FastLED7.5k—~897Automated safety check: PassMIT
Bats Shell Testing Patternswshobson/agents40k12 repos~1.3kAutomated safety check: PassMIT
Ue Test AuthoringJasonMa0012/MooaToon750—~2.1kAutomated safety check: NotesCustom licence
Playwright6BNBN/FlowPilot134—~1kAutomated safety check: PassApache-2.0

Similar skills

  • E2E

    callstack/react-native-pager-view

    Agentic end-to-end tests with e2e, the e2e runner. An agent skill from callstack/react-native-pager-view.

    3.4k GitHub starsUsed in 1 repo~2.1k tokens
    Testing & QAAuto-check passed
  • CI Fix

    FastLED/FastLED

    Scan all CI builds and tests, find failures, fetch error logs, and fix the code.

    7.5k GitHub stars~897 tokensUpdated today
    Testing & QAAuto-check passed
  • Writes unit tests for shell scripts with Bats: error-condition tests, fixtures and mocks, cross-shell checks, parallel runs, helper files and CI integration.

    40k GitHub starsUsed in 12 repos~1.3k tokens
    Testing & QAAuto-check passed
  • Ue Test Authoring

    JasonMa0012/MooaToon

    A skill your agent uses when writing or modifying UE automated tests (Automation, CQTest, Functional, Gauntlet, LowLevel) with Rider MCP available.

    750 GitHub stars~2.1k tokensUpdated 21 days ago
    Testing & QAAuto-check: notes
  • Playwright

    6BNBN/FlowPilot

    A skill your agent uses when the task requires automating a real browser from the terminal in Cursor, including navigation, form filling, snapshots, screenshots, data extraction, and UI-flow…

    134 GitHub stars~1k tokensUpdated 7 mo ago
    Testing & QAAuto-check passed
  • TDD

    FastLED/FastLED

    Guide Test-Driven Development workflow for FastLED. An agent skill from FastLED/FastLED.

    7.5k GitHub stars~848 tokensUpdated today
    Testing & QAAuto-check passed

More from shangyankeji/super-dev

  • Super Dev

    shangyankeji/super-dev

    Activate the Super Dev pipeline inside Codex CLI. An agent skill from shangyankeji/super-dev.

    273 GitHub stars~2.1k tokensUpdated 3 mo ago
    Auto-check passed
  • Super Dev Seeai

    shangyankeji/super-dev

    Activate the Super Dev SEEAI competition mode inside Codex CLI.

    273 GitHub stars~5.8k tokensUpdated 3 mo ago
    Auto-check passed
  • Super Dev

    shangyankeji/super-dev

    Super Dev Codex App/Desktop plugin entry. An agent skill from shangyankeji/super-dev.

    273 GitHub stars~500 tokensUpdated 3 mo ago
    Auto-check passed
  • Super Dev Seeai

    shangyankeji/super-dev

    Super Dev SEEAI Codex App/Desktop competition entry. An agent skill from shangyankeji/super-dev.

    273 GitHub stars~216 tokensUpdated 3 mo ago
    Auto-check passed

Works with

Categories

Questions about Super Dev

What does Super Dev do?

Super Dev pipeline governance for research-first, commercial-grade AI coding delivery. Super Dev is an agent skill from shangyankeji/super-dev.

When should I use Super Dev?

Super Dev fits situations like: testing & QA work in your project.

How do I install Super Dev in Claude Code?

Run `npx skills add shangyankeji/super-dev --skill super-dev -a claude-code`. Or copy the skill folder (.claude/skills/super-dev in shangyankeji/super-dev) into .claude/skills/super-dev in your project. Claude Code loads it when a task matches its description.

How do I install Super Dev in Codex?

Run `npx skills add shangyankeji/super-dev --skill super-dev -a codex`. Or copy the skill folder (.claude/skills/super-dev in shangyankeji/super-dev) into .agents/skills/super-dev in your project. Codex loads it when a task matches its description.

Can I use Super Dev in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add shangyankeji/super-dev --skill super-dev -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/super-dev, .gemini/skills/super-dev, .github/skills/super-dev and .opencode/skills/super-dev in your project.

What does Super Dev need to run?

Going by SKILL.md and its folder, Super Dev needs Python for the scripts in its folder and the command-line tools its instructions call (pip and codex). Our summary lists: Python 3. Its frontmatter pre-approves these tools: Read, Edit, Write, Bash.

Does Super Dev access the network?

SKILL.md contains no URLs. Its commands use pip, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Super Dev safe to install?

Our automated static check of SKILL.md found notes only (pre-approves every shell command (allowed-tools: bash)), nothing it rates as a warning. It is not a guarantee. The check reads SKILL.md only: the scripts in the folder are not scanned, so read them before running anything.

What licence does Super Dev use?

Super Dev is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Super Dev use?

About 2.6k tokens (SKILL.md is roughly 10k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Super Dev?

Skills that share tags, products or a category with Super Dev: E2E (callstack/react-native-pager-view, 3.4k stars), CI Fix (FastLED/FastLED, 7.5k stars), Bats Shell Testing Patterns (wshobson/agents, 40k stars) and Ue Test Authoring (JasonMa0012/MooaToon, 750 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Super Dev?

shangyankeji (a GitHub user) maintains it in shangyankeji/super-dev, which has 273 GitHub stars. The repository holds 5 skills in this directory. The repository was last updated on June 19, 2026.

Source: shangyankeji/super-dev on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.