agentsclimarketplace

Verification loop

Skill codelably/harmony-claude-code/skills/verification-loop

完整的 Claude Code 配置集合(agents, skills, hooks, commands, rules, MCPs)。源自 Anthropic 黑客松获胜者的实战配置,助力中文工程师高效理解与使用 Claude Code。From the repository description

Install
npx -y skills add codelably/harmony-claude-code --skill verification-loop

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

2.6 KB, as published. Nobody here has run it

验证循环技能(Verification Loop Skill)

一个用于 Claude Code 会话的全面验证系统。

何时使用

在以下场景调用此技能(Skill):

  • 完成功能开发或重大代码变更后
  • 创建 PR 之前
  • 当你想确保质量门禁(Quality Gates)通过时
  • 代码重构之后

验证阶段(Verification Phases)

阶段 1:构建验证(Build Verification)

# 检查项目是否可以构建
npm run build 2>&1 | tail -20
# 或者
pnpm build 2>&1 | tail -20

如果构建失败,请停止并修复后再继续。

阶段 2:类型检查(Type Check)

# TypeScript 项目
npx tsc --noEmit 2>&1 | head -30

# Python 项目
pyright . 2>&1 | head -30

报告所有类型错误。在继续之前修复关键错误。

阶段 3:Lint 检查(Lint Check)

# JavaScript/TypeScript
npm run lint 2>&1 | head -30

# Python
ruff check . 2>&1 | head -30

阶段 4:测试套件(Test Suite)

# 运行带有覆盖率报告的测试
npm run test -- --coverage 2>&1 | tail -50

# 检查覆盖率阈值
# 目标:最低 80%

报告内容:

  • 总测试数:X
  • 通过:X
  • 失败:X
  • 覆盖率:X%

阶段 5:安全扫描(Security Scan)

# 检查密钥
grep -rn "sk-" --include="*.ts" --include="*.js" . 2>/dev/null | head -10
grep -rn "api_key" --include="*.ts" --include="*.js" . 2>/dev/null | head -10

# 检查 console.log
grep -rn "console.log" --include="*.ts" --include="*.tsx" src/ 2>/dev/null | head -10

阶段 6:差异审查(Diff Review)

# 显示变更内容
git diff --stat
git diff HEAD~1 --name-only

审查每个变更的文件,确认:

  • 无意间的变更
  • 缺失的错误处理
  • 潜在的边缘情况

输出格式(Output Format)

运行完所有阶段后,生成一份验证报告:

VERIFICATION REPORT
==================

Build:     [PASS/FAIL]
Types:     [PASS/FAIL] (X errors)
Lint:      [PASS/FAIL] (X warnings)
Tests:     [PASS/FAIL] (X/Y passed, Z% coverage)
Security:  [PASS/FAIL] (X issues)
Diff:      [X files changed]

Overall:   [READY/NOT READY] for PR

Issues to Fix:
1. ...
2. ...

持续模式(Continuous Mode)

对于长时间的会话(Session),每 15 分钟或在重大变更后运行一次验证:

设置心理检查点:
- 完成每个函数后
- 完成一个组件后
- 在开始下一个任务之前

运行:/verify

与钩子(Hooks)集成

此技能(Skill)是对 PostToolUse 钩子(Hooks)的补充,但提供了更深层次的验证。 钩子可以立即发现问题;此技能则提供全面的审查。

Keep looking

Skills are one crate of 614,496. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.