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 Issue Fix

liuzhengdongfortest/codestable
911 installs890 stars
Summary

This handles stage 3 of a bug workflow: taking a confirmed root cause and fix plan, implementing the minimal code changes, verifying the fix works, and writing a fix-note.md to close the loop. It enforces strict scope discipline by only touching files declared in the analysis phase and blocking "drive-by refactoring" that would muddy the PR. The skill runs verification checklists for reproduction steps, expected behavior, and regression surfaces, then switches to log-based debugging if the fix doesn't work instead of letting you guess repeatedly. It's opinionated about keeping bug fixes surgical and traceable, which feels restrictive at first but keeps git blame clean and code review focused.

Install to Claude Code

npx -y skills add liuzhengdongfortest/codestable --skill cs-issue-fix --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-issue-fix

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。

根因和方案已经确定(标准路径在 analysis、快速通道在 report 阶段口头确认过),你的活是按方案改代码、验证效果、写下修复记录。

fix 阶段最容易出问题的不是改代码本身,而是改的过程中冒出的"顺手"冲动——顺手优化、顺手重构、顺手加抽象。每项单独看说得通,但合在一个 PR 里让别人分不清"这次到底为了修 bug 改了什么"。

共享路径与命名约定看 .codestable/reference/shared-conventions.md 第 0 节和 cs-issue 的"文件放哪儿"。


两种入口

标准路径(有 analysis)

  1. 方案已确认——读 analysis,确认 doc_type=issue-analysis 且 status=confirmed,第 5 节用户选定了哪个方案
  2. 上下文读全:analysis 全文 + report 全文 + analysis 第 1 节定位的所有代码 + .codestable/attention.md + 沉淀目录搜索:
    • python .codestable/tools/search-yaml.py --dir .codestable/compound --filter doc_type=trick --filter status=active --query "{关键词}"——确认修复方式不违背已有库用法 / 模式
    • 同样命令换 --filter doc_type=explore——确认修复点和已有证据不冲突
  3. 确认起点——告诉用户"我将按方案 X 修改 {文件列表},开始修复",等用户确认才动手

快速通道(无 analysis,从 report 直接触发)

进入这个入口时 AI 在 report 阶段已读过代码并对根因有把握。

  1. 明确陈述根因:"{文件}:{行号} 的 {具体代码} 存在 {问题描述}",让用户确认根因判断准确
  2. 给修复方案——改哪里、怎么改(一两句话,不写完整分析文档)
  3. 等用户明确说"对,就这样改"才动手——不允许"我觉得对,直接改了"
  4. 读 .codestable/attention.md
  5. 补搜沉淀目录——快速通道也要查一遍 compound/(trick + explore),避免误把已知边界条件当新问题

实现期间的约束

只改 analysis 里声明的文件

修复范围来自 analysis 第 5 节"推荐方案"的"影响面"。超出范围的文件——哪怕顺眼——不动。

发现范围外值得改的记一条"顺手发现"不改代码:

> 顺手发现:{文件:行号} {问题简述}。不在本次修复范围,可后续另开 issue。

为什么这么严:顺手改的代码不在分析里,验收对不上,git blame 分不清哪些改动是为这个 bug。

改动最小化

修复只针对根因,不引入新抽象、新接口、新模式。如果发现"要把这个改好得先重构 X"——停下来跟用户确认是否在这个 issue 里做重构,还是拆成独立工作。

为什么:bug 修复天然窄场景,引入新抽象意味着只有这一个使用点支撑——典型过早抽象。

代码质量反射检查

修 bug 看似动作小但 AI 写修复代码一样会漂——大文件再塞特殊处理、大类再加方法、为绕开边界加 if 分支。反射检查见 shared-conventions.md 第 7 节。

issue-fix 比 feature-implement 更谨慎:触发反射信号但结论是"该拆"时默认不在本次 PR 做——按"改动最小化"记成顺手发现。唯一例外是"不拆就没法干净修这个 bug",那停下来跟用户确认"修这个 bug 的前置是 {重构动作},合进来还是拆出去单独做"。

每完成一处改动必须汇报

修复汇报模板见同目录 reference.md,不允许含糊汇报。汇报后停下等用户回复。


验证清单

修复改完后逐项核对:

  • 复现步骤验证——按 report 第 2 节走一遍,问题不再出现
  • 期望行为验证——report 第 3 节"期望行为"现在确实发生
  • 影响面回归——analysis 第 4 节"潜在受害模块"每个走一遍最基本的冒烟路径
  • 前端改动浏览器验证(如涉及)——按 .codestable/attention.md 的硬要求执行,不能只 typecheck
  • 相关测试通过——有测试覆盖到修复区域就跑一遍

修复未生效时:日志调试升级

走完验证清单仍问题复现或行为与期望不符——别在原有猜测上反复试错,切换到日志调试模式重新收集运行时证据。

为什么切换:反复试错本质是猜测在原假设下还有什么可能性,但如果原假设就错了再猜也是绕圈。日志强制看实际运行时数据,往往一眼看出原假设哪里偏了。

日志调试步骤、用户取日志提示词、循环限制见同目录 reference.md。


写 {slug}-fix-note.md

验证通过后在 issue 目录建 {slug}-fix-note.md(位置见 cs-issue 的"文件放哪儿"),记录完整闭环。标准路径模板和快速通道模板都在同目录 reference.md。


退出条件

  • 所有改动文件已提交或列清单
  • 验证清单全部勾选
  • {slug}-fix-note.md 已建并填写完整
  • 没有未处理的"顺手发现"(都进后续 issue 列表)
  • 没有范围外改动(或已和用户确认)
  • 用户明确确认修复完成

收尾提交

按 shared-conventions.md 第 4 节"scoped-commit"规则执行。本阶段:

  • 提交范围:修复代码 + {slug}-fix-note.md + 本次一并更新的 report / analysis
  • 修复闭环后告诉用户"修复验证已完成,{slug}-fix-note.md 已落盘",紧接着问是否需要 commit

退出后

告诉用户:"issue 修复完成,工作流闭环。report + analysis + fix-note 已存档。"

按 shared-conventions.md 第 3 节"issue-fix"收尾推荐顺序各问一句(用户"不用"立即跳过):

  1. 暴露了值得复用的坑点 → "沉淀 learning?(cs-learn)"
  2. 沉淀出长期约束 / 规约 / 技术决定 → "归档决定?(cs-decide)"
  3. 这个 bug 暴露了项目通用的硬约束 / 命令陷阱 / 环境设置(一两行能讲清、CodeStable 技能每次启动都该知道)→ "记到 attention.md?(cs-note)"
  4. 最后问是否代为提交。同意时按收尾提交规则执行

建议:把 issue 目录文件和代码改动放同一次提交方便追溯;"顺手发现"另开 cs-issue-report 处理别塞这个 PR。

修复中发现问题实际是功能缺失(不是 bug)→ 建议另开 cs-feat,别在 issue 工作流里偷偷做新功能。


容易踩的坑

  • 修完没走验证清单就宣告"修好了"
  • 顺手改了 analysis 范围外的代码
  • 修复引入新抽象 / 接口但没停下来确认
  • {slug}-fix-note.md 没建就宣告完成
  • 发现影响面回归有问题但写"轻微影响可忽略"——要修到干净
  • 前端改动只 typecheck 就报通过
  • 用户没明确说"修复完成"就结束
  • 修复未生效继续原假设上反复猜测试错,不切换到日志调试
  • 日志调试结束后没清理临时 log 就提交
  • 收尾时没问用户是否代为 commit
  • 用户没明确同意就 git commit
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 →
Categories
Git & Pull RequestsDebugging
First SeenMay 16, 2026
View on GitHub

Recommended

More Git & Pull Requests →
github-pr-review

fvadicamo/dev-agent-skills

github pr review
491
63
github-pr-merge

fvadicamo/dev-agent-skills

github pr merge
214
63
make-pr-easy-to-review

cursor/plugins

Prepare PRs for review by cleaning noisy history, improving PR descriptions, and adding reviewer guidance without changing code behavior.
2.1k
git-commit

github/awesome-copilot

Standardized git commits using Conventional Commits specification with intelligent diff analysis and message generation.
33.7k
34.3k
pr-review-expert

alirezarezvani/claude-skills

Performs systematic code review of GitHub PRs and GitLab MRs with blast radius analysis, security scanning, breaking change detection,...
534
16.9k
pr-review

microsoft/win-dev-skills

Multi-dimensional review of a PR or feature branch in microsoft/win-dev-skills. Activate on "review my PR / changes / branch", "vet before pushing", "PR review", "is this ready to merge". Fans out parallel sub-agents over skill content, the skill-vs-tool boundary (solution hierarchy), tool correctness, payload/provenance/tests, docs & manifest sync, plus a multi-model cross-check. Reports findings to stdout. Does NOT apply fixes.
288