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.
npx skills add YangFan-Code-Star/context-dev --skill product-designgit clone --depth 1 https://github.com/YangFan-Code-Star/context-devWrote 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.
[](https://agentmods.dev/skills/yangfan-code-star/context-dev/product-design)<a href="https://agentmods.dev/skills/yangfan-code-star/context-dev/product-design"><img src="https://agentmods.dev/badge/skills/yangfan-code-star/context-dev/product-design/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.
<a href="https://agentmods.dev/skills/yangfan-code-star/context-dev/product-design"><img src="https://agentmods.dev/badge/skills/yangfan-code-star/context-dev/product-design.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00084 | $0.01356 |
| Opus 5.5 | $0.00034 | $0.00542 |
| Sonnet 5.5 | $0.00017 | $0.00271 |
| Haiku 4.5 | $0.00008 | $0.00136 |
Grade A, and why
product-design 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 1mo 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.
What it actually says
产品设计方法
唯一信条
产品是「问题被解决」之后自然沉淀的产物,不是起点。
设计的对象从来不是「做什么功能」,而是「回答什么问题」。任何功能开工前必须先改写成契约。
契约:动词 → 三行
每个功能只允许是这三行,超出即臃肿:
| 行 | 问题 |
|---|---|
| 1 | 它回答用户的哪个问题? |
| 2 | 它唯一输出是什么(一个数字 / 一张图 / 一个结果)? |
| 3 | 用户拿这个输出做哪个决定? |
答不出第 3 行 = 这个功能没有存在理由,写进 docs/roadmap.md 的「明确不做」,不写进产品。
形态无关:Web 的一屏、App 的一个页面、CLI 的一个命令、库的一个函数、抽象产品的一个环节,套的是同一个契约。变的只是「输出怎么呈现」,契约本身不变。
原则层
- 单一权威源:契约表是唯一权威,别处不重复定义同一个功能。重复定义 = 漂移的起点。
- 语义分层:界面只引用语义(「结论」「警示」「危险」),不写死样式值;消费方永远不碰原始值。
- 缺失 ≠ 要补:缺一个字段 / 颜色 / 功能,先用最接近的既有答案顶上,而不是加一个新的。补进来的东西要有「设计负责人批准」级别的理由。
- 责任分离:一个界面 / 环节只回答一个问题,不越界干别的。
- 例外走显式流程:想加任何东西,先写进
docs/roadmap.md的「明确不做」区,再走过滤问题决定要不要进产品。
方法层(每次迭代都执行)
外部基准(默认步骤,先于契约确认)
用户提出功能构想时,先做轻量市场基准,再确认契约三行:
- 问一句:同类成熟产品如何回答同一个问题(它的 问题→输出→决定)?国际 / 中文 / 开源各覆盖 1–2 个方向,定向检索。功能名只是检索词——先重写成三行让用户确认,再继续。检索不可用(离线 / 无网络)时,改问用户要 1–2 个他熟悉的同类产品代替;仍没有就跳过基准,并在契约备注里标注「基准未做」——基准是输入不是门禁。
- 收敛成参考:把「别人的 问题→输出→决定」压缩成一段外部基准(附来源),随契约落盘为一行备注。外部阈值与口径不可移植,只借鉴「回答了什么问题、输出了什么、驱动什么决定」的结构。
- 边界:不建研究账本、不做多路并行、不设交付格式门——基准是契约的输入,不是独立项目。
过滤问题(唯一的砍刀)
想加任何功能、字段、图表、按钮、命令、参数,先问一句:
它服务契约表里哪个决定?没有它,那个决定还做得出来吗?
- 答不上 → 写进
docs/roadmap.md的「明确不做」,不进产品。 - 做得出来 → 砍掉,它此刻就是臃肿。
- 做不出来 → 保留,它是刚需。
不需要 RICE、Kano、优先级框架来「排序」——排序是等它进来之后的事,砍刀只需要这一句。
结论前置
默认只显示「触发决定的那个结果」(结论数字),明细、图表、表格全部「点开才有」。臃肿几乎总是「把明细默认铺开」造成的。
显示导致决定的数字,不是原始数据。比如显示「储蓄率 38%」,而不是「年支出 20 万」——前者直接触发决定,后者还要用户心算。
「明确不做」仪式
每次想加东西,先写进 docs/roadmap.md 的「明确不做」区(有地方安放,不憋着),再走过滤问题:确认不要的留在该区;被判定为刚需的,从该区移进对应里程碑的功能表。这份清单同时是「想加的东西有去处」的安放处和「迷茫时的方向标」。
与 init 的协作
init 技能阶段 2(领域深挖)在项目形态是面向用户的产品时,应调用本技能做一次契约访谈,把收敛结果落进 docs/goals.md 和 docs/roadmap.md 的「明确不做」。本技能是独立技能,可单独调用,也可被 init 引用,两者各自演进。
反模式(这些写法等于没写)
- 用「以后可能用到」作为保留理由 → 直接砍。加回来的成本近零,方向错的成本不可逆。
- 为还不存在的需求预建字段 / 目录 / 接口 → 砍。
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.
- 1mo ago First seen · 76 lines · 84 tokens per session scan A 5c3667173c6f
product-design is a skill published in the GitHub repository YangFan-Code-Star/context-dev (2 stars, last pushed 2mo ago), licensed MIT. It adds 84 tokens to every session and 1,356 once invoked, about $0.0003 per session on Opus 5.5. 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 2026-08-31.
Other skills, from other repositories
c4-diagrams
Use when explaining existing code architecture, visualizing a new system before detailed design, mapping software boundaries, or creating C4-style diagrams in ASCII or Mermaid.
figure-drawing
A workflow for making editable draw.io diagrams for research papers, technical systems, concepts, processes, and data.
design-taste-frontend
A frontend-design skill for creating distinctive landing pages, portfolios, and redesigns. It first infers the desired visual character and then proposes a design direction and rules.
product-behavioral-nudge-engine
A guide to using behavioural psychology when shaping interface timing, wording, and incentives.
prototype
A quick, disposable working prototype for testing whether a design idea, state flow, or interface makes sense.
motion-designer
Guidance for designing interface animation, including small interactions, screen changes, loading feedback, gesture responses, timing, and easing. It also considers reduced motion, an accessibility setting for people who prefer less animation.