CAT
/Skills
SkillsMCPMarketplacesDigestToolsAdvertise

This week in Claude

Every Monday: Claude Code, Agent SDK, MCP, and the Anthropic platform moves worth your time.

Skills by Category
Frontend DevelopmentBackend & APIsTesting & QASecurityDevOps & CI/CDGit & Pull RequestsDocumentationCode Review & QualityAI & Agent BuildingSkill Development
MCP Servers by Category
Sales & MarketingWeb & Browser AutomationDatabasesAI & LLM ToolsCloud & InfrastructureCommunication & MessagingDeveloper ToolsDesign & CreativeDocuments & KnowledgeSearch & Web Crawling
Marketplaces by Category
AI Agents & OrchestrationLLM IntegrationDevelopment ToolsFrontend & UIBackend & APIsDatabasesTesting & Code QualityDevOps & CloudSecurity & ComplianceGit & Version Control

Cross AI Tools

Discover Claude Code plugins, extensions, and tools. Automatically updated directory of Anthropic Claude AI marketplaces with development tools, productivity plugins, and integrations.

Resources

  • Browse Skills
  • Browse MCP Servers
  • Browse Marketplaces
  • Plugins Reference

Community

  • About
  • Tools
  • Feedback
  • Privacy Policy
  • Advertise

Built for the Claude Code community with Claude Code by @mertduzgun

Independent project, not affiliated with Anthropic

Cs Feat Ff

liuzhengdongfortest/codestable
906 installs890 stars
Summary

This is the escape hatch for small features that don't need the full design process. When a user says "快速模式" or "直接开干", it skips design docs and checklists but forces you to search the CodeStable knowledge base first (architecture, past decisions, learned lessons) before writing any code. You write the feature, validate it, then drop a minimal ff-note.md with four sections: what you did, files changed, how you verified, things you noticed. It's intentionally lightweight, maybe 15 lines total. The moment you realize you're touching three subsystems or rewriting terminology, you bail out to cs-feat-design. Think of it as "yes, move fast, but at least check what others already figured out first."

Install to Claude Code

npx -y skills add liuzhengdongfortest/codestable --skill cs-feat-ff --agent claude-code

Installs into .claude/skills of the current project.

CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
Keep your Mac awake
Keep your Mac awake
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Context.devContext.dev
Context.dev
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Make money from your Skills
Make money from your Skills
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
Keep your Mac awake
Keep your Mac awake
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Context.devContext.dev
Context.dev
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Make money from your Skills
Make money from your Skills
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
Files
SKILL.mdView on GitHub

cs-feat-ff

启动必读

开始任何判断或动作前,先读取 .codestable/attention.md;缺失则视为骨架不完整,提示先补齐或运行 cs-onboard,不要回退到外部 AI 入口文件。

用户让你做小功能时本来 AI 就会直接动手——这个技能不改变这件事。它只做一件事:动手前把项目里已沉淀的 CodeStable 知识指给你,按需搜一下,写出来的代码就比裸写多一层保护;动手后回写一份最简的 {slug}-ff-note.md 让这次工作可追溯、可被 cs-arch / cs-req backfill 看到、能纳入 scoped-commit 提交。

很轻:没有 design doc / checklist / 验收清单 / 动手前的用户确认。看完指引,该读代码读、该写代码写、写完回写一段话。


动手前先扫一眼 .codestable/

Glob .codestable/ 发现可用目录和文档,按需取用:

  • architecture/ — ARCHITECTURE.md 总入口 + 子系统 doc。改跨模块的东西前看一眼避免违反边界
  • compound/ — learning / trick / decision / explore 四类沉淀:
    python .codestable/tools/search-yaml.py --dir .codestable/compound --filter doc_type=learning --query "关键词"
    python .codestable/tools/search-yaml.py --dir .codestable/compound --filter doc_type=decision --query "关键词"
    python .codestable/tools/search-yaml.py --dir .codestable/compound --filter doc_type=trick --query "关键词"
    
  • requirements/ — 有相关 req 时读边界
  • features/ — 有同类 feature 时参考其 design
  • reference/ — shared-conventions.md / tools.md

怎么用

动手前问 2 个问题:

  1. 这块代码以前有人栽过跟头吗? → 搜 compound/ 的 learning
  2. 这块代码有没有已经拍板的写法约束? → 搜 compound/ 的 decision + 看 architecture/ 相关子系统

命中就把结论融进实现(按约束来写,不是抄)。没命中按自己判断写很正常。搜不到换几个关键词再试。


写代码时守住这几条

design / implement 的硬约束在 fastforward 的精简版。没 design doc 不代表可以不讲——这些是让你"直接动手"时不偏向 AI 默认会踩的坑。

先想"放在哪儿"再写

30 秒回答:这次要加的东西在项目结构里属于哪儿?

  • 现有模块本该承担?→ 在那扩展,别另起
  • 横跨多个模块?→ 抽公共层 / 让某一方主导
  • 已有模块叫法不同?→ grep 同义词
  • 跟现有都不像?→ 多半该切回完整 design 流程

默认坑:不思考就往眼前最顺手的文件里加——加完就成"什么都装的筐"。

扫一眼要改的文件现在什么状况

开写前看一眼:文件多长?承担几件事?类有多少方法?新加的是自然扩展还是把它推向"什么都能干"?

健康就直接加;要先收拾(拆长文件 / 抽重函数)就先收拾再加,范围锁死为"只搬不改行为";结构性问题(职责重划 / 模块拆合)→ 停下来回 design 流程。

默认写最少的代码

只写用户明确要的。不顺手加:

  • "以后可能要"的配置项 / 参数开关 / 抽象层 / 接口 / 工厂
  • 没人要的防御性兜底 / try-catch
  • 用户没提的边界处理

判据:写完觉得"是不是还得加点 X"——X 是不是用户能感知到的?不是就别加。多出来的代码不是中性的,是后人维护的负担。

只动该动的

只改要改的函数。同文件里别的函数丑 / 命名怪——除非和这次冲突,否则别碰。新代码风格匹配当前文件已有写法。看到值得改的别处 → "顺手发现:{文件:行号} {问题},不在本次范围"让用户决定。

新逻辑默认放新文件

会被其他地方引用 → 新文件;只一处用的小工具函数 → 就近放。

不打补丁分支

冒出 if (特殊情况) { 特殊处理 } → 停。这种分支基本只因为思路没覆盖到这种情况,硬写下去得到的是"为让代码能跑而加的特殊逻辑"。要么改数据结构让它不需要特殊处理,要么明确承认是边界情况并注释说明为什么特殊。

反射信号触发就停

  • 往 > 300 行文件追加 / 往 > 10 个方法的类加方法
  • 函数做的事越来越多超过一屏
  • 写第二段"跟上面那段基本一样改了两个变量"的代码
  • 函数参数加到第 4 个
  • 往 utils.ts / helpers.ts 万能 util 堆东西
  • 新起概念名时先 grep 同名 / 近义命名

完整清单看 .codestable/reference/shared-conventions.md 第 7 节。


写完回写 {slug}-ff-note.md

代码写完、验证完、用户确认效果 OK 之后才动这一步——动手前先建空壳会破坏 ff 的轻体感。

自动生成 slug

不问用户。规则:

  • 从用户最初的请求里抽 2-4 个英文 kebab-case 词概括动作 / 对象(如 "rename-cwd-helper"、"add-export-button"、"fix-toolbar-layout")
  • 不抽业务名词缩写、不音译中文、保持小写连字符
  • 拿不准就保守一点:"tweak-{对象}" / "small-{动作}" 也比强求精准好

最终路径:.codestable/features/YYYY-MM-DD-{slug}/{slug}-ff-note.md,日期用今天。

模板

---
doc_type: feature-ff-note
feature: {slug}
date: YYYY-MM-DD
requirement: {req-slug 或留空}
tags: [...]
---

## 做了什么
{1-3 句:解决什么需求 / 加了什么能力,业务视角}

## 改了哪些
- {file:行号区间或函数名} — {一句话说改了什么}
- ...

## 怎么验证的
{1-2 句:跑了哪些验证 / 浏览器走通了哪条路径 / 跑了什么测试}

## 顺手发现(可选,不阻塞)
- {文件:行号} {问题简述} — 不在本次范围

写得真的轻:每节就那么几行,不要把它写成迷你 design / 迷你 acceptance。这份文档的目标是"半年后有人看 git log 能跳进来 30 秒搞清楚做了啥",不是替代标准流程。

落盘后告诉用户:"已写 {slug}-ff-note.md,本次 fastforward 闭环。"


不做什么

  • 不写 design doc / checklist / acceptance——这就是 fastforward 的意义。要写就去 cs-feat-design
  • 不跟用户确认方案——用户让你做小功能就是不想等你开会
  • 不在 .codestable/ 里留 {slug}-ff-note.md 之外的新文件——除非发现值得沉淀的坑 / 技巧,另起对话用 cs-learn / cs-trick 写

什么时候跳出 fastforward

干到一半发现下面任一情况,停下来告诉用户"这比想象的复杂,建议切回完整流程":

  • 改动涉及 3 个以上子系统
  • 需要引入新术语或和现有术语冲突
  • 要动 .codestable/architecture/ 既定的模块边界
  • 用户追加的要求让范围翻倍

切回方式:触发 cs-feat-design。已写的代码在 design 里标"已部分实现"即可。


退出条件

  • 代码写完且用户确认效果 OK
  • {slug}-ff-note.md 已落盘且四节填齐(顺手发现可省)
  • 没有未对齐的"顺手发现"(都进 ff-note 末节,留给后续)

收尾提交

按 .codestable/reference/shared-conventions.md 第 4 节"scoped-commit"规则执行。本通道:

  • 提交范围:本次代码改动 + {slug}-ff-note.md
  • ff-note 落盘后告诉用户"已就绪,是否代为 commit?",用户明确同意才执行

按 shared-conventions.md 第 3 节"feature-ff"收尾推荐顺序逐项一句话提示(用户"不用"立即跳过):

  1. 暴露的坑 → "沉淀 learning?(cs-learn)"
  2. 拍板的长期约束 → "归档决定?(cs-decide)"
  3. 最后问是否代为 scoped-commit

容易踩的坑

  • 完全跳过知识检索就写——这个技能的唯一理由就是让你搜一下再写
  • 把搜到的 learning / decision 当"参考"而不是"约束"——decision 拍过板,违反要么重新 decision 要么别做
  • 开始写 design doc——fastforward 就是不写 design
  • 发现任务变复杂还硬在 fastforward 推——切回成本远低于带着错误方案改到底
  • 动手前就建 ff-note 空壳——破坏 fastforward 的轻体感,必须代码 + 验证完才回写
  • 把 ff-note 写成迷你 design / 迷你 acceptance——四节加一起十几行就够,多了说明这事不该走 fastforward
  • 跳过 ff-note 直接 commit——和 issue-ff 强制 fix-note 同样的理由:没记录后人追溯不了
Featured
CodeRabbit
CodeRabbit
AI writes the code. CodeRabbit catches the slop.
Try For Free →
Keep your Mac awake
Keep your Mac awake
Keep your Mac awake while Claude Code and 40+ AI agents run. Sleeps when they're idle.
One time payment $9 →
Context.devContext.dev
Context.dev
Integrate web data into your AI product. One API to scrape website & brand data.
Get API Key Now →
Make your agent a DeFi expert
Make your agent a DeFi expert
Agent, run crypto. Access onchain data & trade routes via 1inch.
Install now →
Make money from your Skills
Make money from your Skills
On Capafy, your Skill runs online 24/7 as an agent product, and you get paid every time someone uses it.
Start earning →
AppSignal
AppSignal
Monitor with ease. Code with confidence.
Start Free Trial →
First SeenMay 16, 2026
View on GitHub

Recommended

caveman

juliusbrussee/caveman

Ultra-compressed communication mode cutting token usage ~75% while preserving technical accuracy.
203.4k
67.8k
grill-me

mattpocock/skills

Relentless interviewing skill that stress-tests plans and designs through systematic questioning.
250.9k
114.5k
improve

shadcn/improve

Survey any codebase as a senior advisor and produce prioritized, self-contained implementation plans for other models/agents to execute.
10
205
systematic-debugging

obra/superpowers

Structured debugging methodology that mandates root cause investigation before attempting any fixes.
124.6k
215.9k
karpathy-guidelines

forrestchang/andrej-karpathy-skills

Behavioral guidelines to reduce common LLM coding mistakes through explicit assumptions, simplicity, and verifiable success criteria.
13.9k
165.4k
find-skills

vercel-labs/skills

Discover and install specialized agent skills from the open ecosystem when users need extended capabilities.
1.8M
21.1k