skill-designer

skill-designer is a skill for Claude Code, Codex from TashanGKD/tashan-cursor-skills. It costs 123 tokens per session (5,288 once invoked), scanned A, original, MIT.

A design role for creating or restructuring Skills, agents, and rules through a guided process. It checks both what the human designer intends and how an AI might interpret the instructions.

In plain words
What is it for?
Use it to design new Skills, agents, or rules, choose the appropriate type, and validate their behavior through staged reviews.
Why use it?
It helps expose ambiguous instructions, missing safeguards, and mismatches between intended behavior and actual execution before deployment.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one.

Good fit Use it to design new Skills, agents, or rules, choose the appropriate type, and validate their behavior through staged reviews.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/tashangkd/tashan-cursor-skills/skill-designer
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add TashanGKD/tashan-cursor-skills --skill skill-designer
Clone the repo
git clone --depth 1 https://github.com/TashanGKD/tashan-cursor-skills

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for skill-designer

README.md
[![agentmods](https://agentmods.dev/badge/skills/tashangkd/tashan-cursor-skills/skill-designer/github.svg)](https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/skill-designer)
Your own site
<a href="https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/skill-designer"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-cursor-skills/skill-designer/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for skill-designer

Your own site · 80×15
<a href="https://agentmods.dev/skills/tashangkd/tashan-cursor-skills/skill-designer"><img src="https://agentmods.dev/badge/skills/tashangkd/tashan-cursor-skills/skill-designer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 123 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 5,288 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00123 $0.05288
Opus 5 $0.00062 $0.02644
Sonnet 5 $0.00025 $0.01058
Haiku 4.5 $0.00012 $0.00529

Measured 7d ago against content hash 00c7ad05af1e, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-11, from the pricing page.

Security

Grade A, and why

skill-designer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 7d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/skill-designer/SKILL.md · 354 lines

How it starts

The opening of the file, as written. The whole thing — 354 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Skill 设计师(skill-designer)

Skill 是双用户产品:人类设计者 + AI 执行者。 这两层有天然张力——对人清晰的描述,AI 可能执行成不同的路径。 本 Skill 的核心价值:在设计过程中,同时从两个视角审视每一个决策。


知识导航表(激活前按 D0→①②③④ 顺序读取)

层级 文档 用途
D0 认知根确认 _内部总控/认知结构/L1_系统性文档/系统架构思维维度/Skill体系设计原则_v1.0.md§2.5 三型决策树 + §2.2/2.4 + §4.3.5 + §一 先于一切,必须全读:① §2.5 统一决策树确定应建 Rule/Agent/Skill(三型判断);② §4.3.5 认知根原则;③ §一唯一最终原则。带三型判断结论进入 Step 1。
① Skill 索引 .cursor/skills/skill-index/SKILL-INDEX.md 获取现有组件列表,判断复杂度级别
② 产品定义模板 .cursor/skills/skill-designer/skill-product-definition-template.md Skill 产品定义卡片标准格式
③ 元项目顶层(可选) _内部总控/元项目导航.md 若新 Skill 与某个子项目强相关,确认其边界约束

激活后立即执行

Step 1  了解需求 + 三型确认(⚠️ 必须按 §2.5 决策树引导,不可让用户自行猜测类型)

        用 D0 读取的 §2.5 决策树,主动引导用户完成三型判断:

        问题 A(对应决策树①):
        「这个需求是「普遍约束」(所有任务都必须遵守)且轻量,还是「执行某类任务的流程」?」
        → 全局约束 → 建 Rule,走 skill-rule-修改规范(告知用户,本 Skill 不继续)
        → 执行流程 → 继续问题 B

        问题 B(对应决策树②):
        「这个任务需要「独立视角隔离」(审核/评测)或「并行批量执行」吗?」
        → 需要隔离/并行 → 建 Agent,走 agent-io-contract + skill-rule-修改规范
        → 不需要 → 建 Skill,继续 Step 2

        ⚠️ 禁止:跳过此引导,让用户自己说「我要建 Skill/Agent/Rule」——用户不一定知道区别。

Step 2  判断复杂度级别
        Read: .cursor/skills/skill-index/SKILL-INDEX.md(获取现有组件列表)
        
        Read: _内部总控/认知结构/L1_系统性文档/系统架构思维维度/Skill体系设计原则_v1.0.md
        → 按知识导航表 D0 行读取(§4.3.5 认知根原则 + §一唯一最终原则)
        → 带着问题进入 Step 3:「本次设计的 Skill,认知根是什么?」
          (认知根 = 这个 Skill 执行的工作,对应认知结构中哪个 L1/L1.5 文档)
        
        根据以下标准判断:
        
        Level 1 补丁型:修改现有组件的一个步骤/注意事项/触发词
          → 无需本 Skill 继续,直接走 skill-rule-修改规范 流程
          → 告知用户:「这是 Level 1 修改,直接按修改规范执行即可」
        
        Level 2 新增型:新建独立组件,与现有系统无显著交互
          → 走 Step 3-6(含关卡A + 关卡C)
        
        Level 3 集成型:新建组件,与多个现有组件有依赖或影响关系
          → 走 Step 3-7(含关卡A + 关卡B + 关卡C)
        
        Level 4 系统型:重构多个现有组件的交互关系
          → 走完整 Step 3-8,并在 Step 3 后暂停等待用户二次确认规模

Step 3  双视角产品定义(引导完成)
        Read: .cursor/skills/skill-designer/skill-product-definition-template.md
        
        按模板逐节引导用户填写,每节结束时切换视角验证:
        
        【人层设计完成后,立即做 Agent 层翻译】
        你(郑总)刚才说的意图,如果 AI 只看文字,会怎么理解?
        我来帮你找出「人觉得显然,但 AI 需要被显式告知」的部分。
        
        对每个触发词:
        → 追问:「在什么相似但不同的情况下,这个词也可能出现?」
        
        对每个步骤:
        → 追问:「如果 AI 不确定,它会倾向于跳过这步还是乱猜?」
        
        对每个输出:
        → 追问:「输出的格式和存放位置是否完全明确?」
        
        【额外必问:失败模式(2.5节)】
        → 追问:「这个 Skill 最容易在哪里出问题?
          过去见过的类似任务,失败原因是什么?
          如果 AI 只走到一半就自以为完成了,会在哪个步骤停下?」
        
        【额外必问:Rule 绑定(2.6节)】
        → 追问:「这个 Skill 处理的是高风险知识任务吗?
          如果是,应绑定 R1 EVIDENCE_FIRST / R2 NO_FABRICATION / R10 MEMORY_BOUNDARY_EXPLICIT。
          它是否依赖正式工件输出?若是,绑定 R6 ARTIFACT_FIRST。
          它是否涉及版本管理?若是,绑定 R9 VERSION_CONTINUITY。」
        
        【额外必问:能力依赖声明(2.7节)——防止能力层重复,最容易漏的一步】
        
        先执行强制扫描(不允许凭记忆回答):
          Read: .cursor/skills/skill-index/SKILL-INDEX.md
          → 找到「能力层 Skill」分类(含 ai-image-generator、write-task-log 等)
          → 找到「配置真源」文件:_内部总控/凭证/
        
        然后追问:
        「① 这个 Skill 需要调用哪些外部 API、执行文件操作、或调用复杂工具?
          列出每一项,对照能力层 Skill 分类,判断是否已有现成 Skill。
          如果有 → 本 Skill 引用现有能力层 Skill,不得自行内嵌实现。
          如果没有 → 是否需要先建立能力层 Skill?(若被多个 Skill 调用,必须先建)
          
         ② 这个 Skill 编写完成后,现有哪些 Skills 应该更新为引用本 Skill?
           (若本 Skill 是能力层,列出所有可能的上层消费者)
           
         ③ 如果涉及 API Key / 模型名 / Endpoint URL,是否已在
           _内部总控/凭证/ 有对应配置文件?
           有 → 引用,不硬编码
           没有 → 先创建配置文件,再引用」
        
        ⛔ 禁止:在流程层/角色层 Skill 里硬编码 API Key、模型名、Endpoint
        ⛔ 禁止:用「我不确定有没有现成 Skill」来跳过扫描,必须真正查 SKILL-INDEX
        
        产出:填写完整的 Skill 产品定义卡片(含 2.5 失败模式 + 2.6 Rule 绑定 + **2.7 认知根文档**)
        保存到:.cursor/skills/skill-designer/draft-[组件名].md
        
        ⚠️ 2.7 认知根文档(新增必填字段,来自 Skill体系设计原则_v1.0.md §4.3.5):
        「本 Skill 执行的工作,对应认知结构中的哪个 L1/L1.5 文档?」
        → 若能指出 → 填写文档路径 + 一句话说明关联关系
        → 若无认知根 → 显式标注「暂无认知根,后续补充」,不允许静默跳过

Step 4  编写 Skill/Agent/Rule
        基于产品定义卡片,编写实际文件
        原则:
        - 人层段落(description、注意事项):面向人类,清晰可读
        - Agent 层段落(执行步骤):面向 AI,精确无歧义,每步有输入/输出/失败处理
        - 两层在同一文件中,用 > [人层说明] 和步骤列表区分

Step 4.5  【F-022 全节点挑战者反思】Skill/Rule 初稿完成后、关卡A前执行
          以「刚拿到这个 Skill 但毫无背景知识的 AI 执行者」视角执行3条挑战:
          
          1. 歧义点:初稿中哪3个词或步骤,AI 执行者最可能理解成与设计意图不同的含义?
             每个触发词在什么相似场景下会被误触发?
          2. 冲突检测:这个 Skill/Rule 的触发词、步骤、输出,与现有哪个 Skill/Rule 有最大的
             潜在冲突?如果两者同时被触发,AI 会怎么选择?
          3. 半途而废:AI 执行者最可能在哪个步骤「自以为完成」而实际上跳过了后续关键步骤?
             这个步骤是否已经有足够明确的「继续/完成」判断标准?
          
          若发现可修复的问题 → 修改初稿后再进关卡A
          若确实无重大问题 → 输出「Skill自检:[歧义/冲突/截断点的具体位置]」

Step 5  关卡A:调用 skill-simulator 子智能体(Level 2+)
        调用 /skill-simulator,传入:
        - 新编写的组件完整内容
        - SKILL-INDEX.md 的当前摘要(现有组件列表)
        
        等待结果,若发现 🔴 严重歧义:
        → 回到 Step 4 修订
        → 再次调用 skill-simulator 验证
        → 直到关卡A通过

Step 6  关卡B:调用 skill-system-destroyer 子智能体(Level 3+)
        调用 /skill-system-destroyer,传入:
        - 新编写的组件完整内容
        - SKILL-INDEX.md 当前摘要(特别是能力层分类)
        - 所有 alwaysApply Rule 的内容
        
        **关卡B必须额外检查(能力重复维度)**:
        → 新 Skill 是否引入了本应已存在于能力层 Skill 的代码?
        → 新 Skill 是否硬编码了 API Key / 模型名 / Endpoint?
        → 如果本 Skill 是流程层/角色层,它调用的所有能力是否都已有对应能力层 Skill?
        
        等待结果,若发现 🔴 Critical:
        → 回到 Step 4 修订
        → 再次调用 skill-system-destroyer 验证
        → 直到关卡B通过

Step 7  关卡C:调用 verifier 子智能体(Level 2+)
        调用 /verifier,传入:
        - 组件文件路径
        - 正向验证场景(「没有这个 Skill,AI 会犯错」的场景描述)
        - 负向验证场景(「这个 Skill 不应该触发」的场景描述)
        
        等待结果,若未通过:
        → 回到 Step 4 修订

Step 8  部署(所有级别)
        调用 skill-rule-修改规范 的 Step 4-6:
        - 备份(如修改已有文件)
        - 执行最小化写入
        - 追加变更记录
        - 更新 SKILL-INDEX.md
        - 更新 PENDING-SKILLS.md(移动到「已完成」区)
        - 更新 role-menu.mdc(强制步骤,SK-001 Gap 修复):
            所有新建 Skill 必须在 role-menu.mdc 的「可用角色一览」表格中注册,
            并在「使用方式」区块中添加对应的触发词示例。
            不允许「条件性」跳过(原措辞「如新 Skill 需要」导致此步骤常被遗漏)。
            若新 Skill 只作为子任务被调用(用户不直接触发),
            可在 role-menu 注释区域标注「内部调用,无用户触发词」。
        
        - 经验沉淀触发(G-01 修复,大闭环接通):
            部署完成后,回顾本次 Skill 设计过程,判断是否有以下信号:
            · 关卡A/B/C 发现了某类反复出现的设计缺陷模式(踩坑)
            · Step 3 双视角引导中遇到了文档未覆盖的新设计决策场景(新发现)
            · 某个追问步骤在实际执行中经常被跳过(步骤偏差)
            IF 有信号 → 追加一行到 .cursor/skills/skill-index/PENDING-EXPERIENCES.md:
              `| [今日日期] | skill-designer | [信号类型] | [一句话描述] | 🔲 待处理 |`
            IF 无信号 → 跳过,不写入

Read the full file on GitHub · 354 lines

Files

What ships with it

31 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 7d ago First seen · 354 lines · 123 tokens per session scan A 00c7ad05af1e

Subscribe to this mod's changes

skill-designer is a skill published in the GitHub repository TashanGKD/tashan-cursor-skills (20 stars, last pushed 5mo ago), licensed MIT. It adds 123 tokens to every session and 5,288 once invoked, about $0.0006 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.

Related

Other skills, from other repositories

template-helpers

Catalog of the reusable building blocks shipped with this Android template — BaseActivity/BaseFragment, the helpers/ extension files (navigation, lifecycle, toast, snackbar, dialogs, images, permissions, theme, locale, settings intents, date, delay), common/ (Firebase, network, observers), and…

orbitalsonic/AndroidPilot · 92 tokens

android-feature

Workflow for implementing a new Android feature, screen, fragment, activity, dialog, adapter, or enhancement in this XML/MVVM template. Use whenever the user asks to add, build, implement, or extend functionality — before writing any code. Covers planning, the MVVM/MVI skeleton, the Android configuration checklist…

orbitalsonic/AndroidPilot · 72 tokens

android-new-project

Workflow for turning a fresh copy of this Android template into a new app — renaming the package/namespace/applicationId, branding (icons, splash, palette, fonts), Firebase setup, signing config, stripping unused sample screens, and verifying the foundation. Use when the user says they are starting a new app…

orbitalsonic/AndroidPilot · 83 tokens

android-preflight

Final verification checklist to run before declaring Android work finished — build, both themes, string resources, lifecycle and leak risks, registered permissions and components, resource parity between values and values-night, and honest reporting of what was and was not verified. Use at the end of any feature, fix…

orbitalsonic/AndroidPilot · 79 tokens

android-upgrade

Workflow for changing an existing working Android project — upgrading dependencies, AGP/Gradle or SDK levels, migrating deprecated APIs, refactoring, and fixing bugs, without breaking what already works. Covers baseline verification, scoping the diff, targetSdk behaviour changes, SharedPreferences/Room data…

orbitalsonic/AndroidPilot · 86 tokens

android-build-config

Gradle and build configuration rules for this template — AGP 9 / Gradle 9 DSL (built-in Kotlin, compileSdk block, optimization block instead of minify/proguard), the gradle/libs.versions.toml version catalog, R8 keep rules in keepRules/.keep, gradle.properties defaults, and manifest registration. Use when editing…

orbitalsonic/AndroidPilot · 104 tokens