Back to Details

legacy-to-ai-ready

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Legacy to AI-Ready

从遗留项目到AI就绪

Transform legacy codebases into AI-ready projects by generating Claude Code configurations.
借助Claude Code配置将遗留代码库转换为AI就绪项目。

Quick Start (5-Minute Setup)

快速开始（5分钟设置）

For most projects, start with just CLAUDE.md:
  1. Analyze: 
    python scripts/analyze_codebase.py [path]
  2. Create CLAUDE.md with build commands, code style, architecture overview
  3. Done - Claude can now write code following your project's conventions
Expand to full configuration only when needed.
对于大多数项目，只需从CLAUDE.md开始：
  1. 分析：
    python scripts/analyze_codebase.py [path]
  2. 创建CLAUDE.md，包含构建命令、代码风格、架构概述
  3. 完成 - Claude现在可以遵循你的项目约定编写代码
仅在需要时再扩展为完整配置。

Interactive Discovery

交互式探索

Before generating configs, ask these questions:
Project Scope:
  • What is this project? (web app, API, CLI, library)
  • Primary language and frameworks?
  • Team size? (solo, small team, enterprise)
Pain Points:
  • What mistakes do new developers commonly make?
  • What patterns should always be followed?
  • What operations are repeated frequently?
Integration Needs:
  • External services used? (databases, APIs, cloud)
  • CI/CD pipeline? (GitHub Actions, Jenkins)
  • Code quality tools? (linters, formatters)
生成配置前，请先询问以下问题：
项目范围：
  • 这是什么类型的项目？（Web应用、API、CLI、库）
  • 主要语言和框架是什么？
  • 团队规模？（单人、小型团队、企业）
痛点：
  • 新开发人员通常会犯哪些错误？
  • 哪些模式必须始终遵循？
  • 哪些操作会重复执行？
集成需求：
  • 使用了哪些外部服务？（数据库、API、云服务）
  • 是否有CI/CD流水线？（GitHub Actions、Jenkins）
  • 使用了哪些代码质量工具？（代码检查器、格式化工具）

Configuration Decision Tree

配置决策树

Start
│
├─ Small project / Solo dev
│  └─ CLAUDE.md only
│
├─ Team project
│  ├─ Multi-language? → Add .claude/rules/
│  ├─ Complex domain? → Add .claude/skills/
│  ├─ Code reviews? → Add .claude/agents/
│  └─ Repeated tasks? → Add .claude/commands/
│
└─ Enterprise / Large team
   └─ All configurations + MCP servers + Hooks
Start
│
├─ Small project / Solo dev
│  └─ CLAUDE.md only
│
├─ Team project
│  ├─ Multi-language? → Add .claude/rules/
│  ├─ Complex domain? → Add .claude/skills/
│  ├─ Code reviews? → Add .claude/agents/
│  └─ Repeated tasks? → Add .claude/commands/
│
└─ Enterprise / Large team
   └─ All configurations + MCP servers + Hooks

Generated Configurations

生成的配置

ConfigPurposeWhen to Create
CLAUDE.mdProject memory (shared)Always (required)
CLAUDE.local.mdPersonal preferences (git-ignored)Individual customization
.claudeignoreFiles Claude should not accessSensitive files exist
.claude/rules/Path-specific rulesMulti-module projects
.claude/skills/Domain knowledgeComplex business logic
.claude/agents/Task specialistsRepeated review/debug tasks
.claude/commands/Quick promptsCommon workflows
.claude/settings.jsonHooks + permissionsAuto-formatting, security
MCP serversExternal toolsDatabase/API integrations
配置用途创建时机
CLAUDE.md项目共享记忆始终需要（必填）
CLAUDE.local.md个人偏好（Git忽略）个性化定制时
.claudeignoreClaude不应访问的文件存在敏感文件时
.claude/rules/路径特定规则多模块项目时
.claude/skills/领域知识存在复杂业务逻辑时
.claude/agents/任务专家存在重复的评审/调试任务时
.claude/commands/快速提示存在常见工作流时
.claude/settings.json钩子与权限需要自动格式化、安全控制时
MCP servers外部工具集成需要数据库/API集成时

Workflow

工作流

Phase 1: Automated Analysis

阶段1：自动化分析

bash
python scripts/analyze_codebase.py [project-path]
The script detects:
  • Languages and frameworks
  • Directory structure patterns
  • Development tools (linters, formatters)
  • Git commit patterns
  • Environment variables
  • Code style indicators
  • CI/CD configurations (GitHub Actions, etc.)
  • Sensitive files (warns about .env, credentials)
Output includes:
  • Recommendations for which configs to create
  • Security warnings for sensitive files
  • Suggested .claudeignore patterns
bash
python scripts/analyze_codebase.py [project-path]
该脚本会检测：
  • 语言和框架
  • 目录结构模式
  • 开发工具（代码检查器、格式化工具）
  • Git提交模式
  • 环境变量
  • 代码风格指标
  • CI/CD配置（如GitHub Actions）
  • 敏感文件（会对.env、凭证文件发出警告）
输出内容包括：
  • 建议创建的配置类型
  • 敏感文件的安全警告
  • 推荐的.claudeignore模式

Phase 2: Context Gathering

阶段2：上下文收集

Claude should read:
  1. 3-5 representative source files - understand naming, patterns
  2. Test files - understand testing approach
  3. Config files - package.json, tsconfig, etc.
  4. README/docs - project overview
  5. Recent commits - understand commit style
Claude应读取以下内容：
  1. 3-5个有代表性的源文件 - 理解命名规则和模式
  2. 测试文件 - 了解测试方法
  3. 配置文件 - package.json、tsconfig等
  4. README/文档 - 项目概述
  5. 近期提交记录 - 了解提交风格

Phase 2.5: Discover Existing Resources

阶段2.5：发现现有资源

Before creating custom configs, search for existing skills and MCP servers:
  1. Search skill marketplaces - SkillsMP, SkillHub.club, Claude Skills Hub
  2. Check GitHub repositories - awesome-claude-skills, themed skill collections
  3. Find relevant MCP servers - Glama, MCP Market, official registry
See references/resource-discovery.md for complete directory of sources.
Tip: Many common needs (git commit, code review, database patterns) already have well-maintained skills available.
创建自定义配置前，请先搜索现有的skills和MCP服务器：
  1. 搜索技能市场 - SkillsMP、SkillHub.club、Claude Skills Hub
  2. 查看GitHub仓库 - awesome-claude-skills、主题化技能集合
  3. 查找相关MCP服务器 - Glama、MCP Market、官方注册表
完整的资源目录请参见references/resource-discovery.md。
提示：许多常见需求（Git提交、代码评审、数据库模式）都已有维护良好的可用skills。

Phase 3: Generate Configurations

阶段3：生成配置

1. CLAUDE.md (Required)

1. CLAUDE.md（必填）

Create at project root with:
  • Quick reference commands (build, test, lint)
  • Naming conventions
  • Architecture overview
  • Testing guidelines
  • Git workflow
See references/claude-md-patterns.md.
在项目根目录创建，包含：
  • 快速参考命令（构建、测试、代码检查）
  • 命名约定
  • 架构概述
  • 测试指南
  • Git工作流
请参见references/claude-md-patterns.md。

2. Rules (If Multi-Module)

2. 规则（多模块项目时）

Create 
.claude/rules/
when:
  • Multiple languages need different conventions
  • Modules have distinct patterns (frontend/backend)
  • Path-specific requirements exist
See references/rules-patterns.md.
在以下情况创建
.claude/rules/
：
  • 多种语言需要不同的约定
  • 模块有不同的模式（前端/后端）
  • 存在路径特定的要求
请参见references/rules-patterns.md。

3. Skills (If Complex Domain)

3. Skills（复杂领域时）

Create 
.claude/skills/
when:
  • Domain-specific workflows exist (database, API patterns)
  • Team knowledge needs preservation
  • Complex procedures are repeated
See references/skills-patterns.md.
在以下情况创建
.claude/skills/
：
  • 存在领域特定工作流（数据库、API模式）
  • 需要保留团队知识
  • 存在重复的复杂流程
请参见references/skills-patterns.md。

4. Subagents (If Specialized Tasks)

4. 子代理（专业化任务时）

Create 
.claude/agents/
for:
  • Code review automation
  • Debugging assistance
  • Security auditing
  • Documentation generation
See references/agents-patterns.md.
为以下场景创建
.claude/agents/
：
  • 代码评审自动化
  • 调试辅助
  • 安全审计
  • 文档生成
请参见references/agents-patterns.md。

5. Commands (If Common Operations)

5. 命令（常见操作时）

Create 
.claude/commands/
for:
  • Git commit workflow
  • PR review process
  • Deployment steps
  • Test running
See references/commands-patterns.md.
为以下场景创建
.claude/commands/
：
  • Git提交工作流
  • PR评审流程
  • 部署步骤
  • 测试运行
请参见references/commands-patterns.md。

6. Hooks (If Auto-Formatting Needed)

6. 钩子（需要自动格式化时）

Configure 
.claude/settings.json
for:
  • Auto-format on file edit
  • Protected files
  • Command logging
See references/hooks-patterns.md.
配置
.claude/settings.json
用于：
  • 文件编辑时自动格式化
  • 受保护文件设置
  • 命令日志
请参见references/hooks-patterns.md。

7. MCP Servers (If External Integrations)

7. MCP服务器（需要外部集成时）

Configure MCP for:
  • Database access
  • GitHub integration
  • Slack notifications
  • Custom internal tools
See references/mcp-patterns.md.
配置MCP用于：
  • 数据库访问
  • GitHub集成
  • Slack通知
  • 自定义内部工具
请参见references/mcp-patterns.md。

Phase 4: Validate

阶段4：验证

  1. Ask Claude to perform a typical task
  2. Verify it follows project conventions
  3. Iterate based on gaps discovered
  1. 让Claude执行一项典型任务
  2. 验证它是否遵循项目约定
  3. 根据发现的不足进行迭代

Output Structure

输出结构

Minimal (small projects):
project/
├── CLAUDE.md
└── [existing files]
Standard (team projects):
project/
├── CLAUDE.md
├── .claude/
│   ├── rules/
│   │   └── code-style.md
│   └── commands/
│       └── commit.md
└── [existing files]
Complete (enterprise):
project/
├── CLAUDE.md              # Shared project memory
├── CLAUDE.local.md        # Personal (git-ignored)
├── .claudeignore          # Files to protect
├── .claude/
│   ├── settings.json      # Hooks + permissions
│   ├── rules/
│   ├── skills/
│   ├── agents/
│   └── commands/
└── [existing files]
最小配置（小型项目）：
project/
├── CLAUDE.md
└── [existing files]
标准配置（团队项目）：
project/
├── CLAUDE.md
├── .claude/
│   ├── rules/
│   │   └── code-style.md
│   └── commands/
│       └── commit.md
└── [existing files]
完整配置（企业级）：
project/
├── CLAUDE.md              # 共享项目记忆
├── CLAUDE.local.md        # 个人配置（Git忽略）
├── .claudeignore          # 受保护文件列表
├── .claude/
│   ├── settings.json      # 钩子与权限
│   ├── rules/
│   ├── skills/
│   ├── agents/
│   └── commands/
└── [existing files]

Reference Materials

参考资料

ReferenceWhen to Read
examples.mdComplete real-world examples
resource-discovery.mdFind existing skills & MCP servers
advanced-patterns.mdMigrations, team collab, monorepos
claude-md-patterns.mdCreating CLAUDE.md
rules-patterns.mdModule-specific rules
skills-patterns.mdDomain knowledge
agents-patterns.mdTask specialists
commands-patterns.mdQuick prompts
hooks-patterns.mdAuto-formatting
mcp-patterns.mdExternal tools
参考文档阅读时机
examples.md完整的真实世界示例
resource-discovery.md查找现有skills和MCP服务器
advanced-patterns.md迁移、团队协作、单体仓库
claude-md-patterns.md创建CLAUDE.md时
rules-patterns.md配置模块特定规则时
skills-patterns.md配置领域知识时
agents-patterns.md配置任务专家时
commands-patterns.md配置快速提示时
hooks-patterns.md配置自动格式化时
mcp-patterns.md配置外部工具集成时

Templates & Bundled Skills

模板与预打包Skills

Templates:
  • assets/CLAUDE.md.template
    - Project memory template
  • assets/settings.json.template
    - Hooks configuration
  • assets/claudeignore.template
    - File ignore patterns
Bundled skills to install in target project:
  • assets/skill-creator/
    - For creating new project-specific skills
  • assets/skill-downloader/
    - For downloading additional skills
  • assets/resource-scout/
    - For discovering existing skills & MCP servers
模板：
  • assets/CLAUDE.md.template
    - 项目记忆模板
  • assets/settings.json.template
    - 钩子配置模板
  • assets/claudeignore.template
    - 文件忽略模式模板
可安装到目标项目的预打包Skills：
  • assets/skill-creator/
    - 用于创建新的项目特定skills
  • assets/skill-downloader/
    - 用于下载额外的skills
  • assets/resource-scout/
    - 用于发现现有skills和MCP服务器

Installing Bundled Skills

安装预打包Skills

Copy these skills to the target project's 
.claude/skills/
directory:
bash
cp -r assets/skill-creator [target-project]/.claude/skills/
cp -r assets/skill-downloader [target-project]/.claude/skills/
cp -r assets/resource-scout [target-project]/.claude/skills/
This enables the target project to:
  1. resource-scout - Discover existing skills & MCP servers before building custom
  2. skill-downloader - Download and install skills from GitHub or archives
  3. skill-creator - Create custom skills tailored to their domain
将这些Skills复制到目标项目的
.claude/skills/
目录：
bash
cp -r assets/skill-creator [target-project]/.claude/skills/
cp -r assets/skill-downloader [target-project]/.claude/skills/
cp -r assets/resource-scout [target-project]/.claude/skills/
这将使目标项目具备以下能力：
  1. resource-scout - 在构建自定义资源前发现现有skills和MCP服务器
  2. skill-downloader - 从GitHub或归档文件下载并安装skills
  3. skill-creator - 创建适合其领域的自定义skills

Language Quick Reference

语言快速参考

TypeScript/JavaScript

TypeScript/JavaScript

  • Extract: eslint, prettier, tsconfig
  • Hooks: prettier auto-format
  • Skills: API patterns, component patterns
  • 提取内容：eslint、prettier、tsconfig
  • 钩子：prettier自动格式化
  • Skills：API模式、组件模式

Python

Python

  • Extract: black, ruff, mypy, pyproject.toml
  • Hooks: black/ruff auto-format
  • Skills: API patterns, ORM patterns
  • 提取内容：black、ruff、mypy、pyproject.toml
  • 钩子：black/ruff自动格式化
  • Skills：API模式、ORM模式

Go

Go

  • Extract: gofmt, golangci-lint
  • Hooks: gofmt auto-format
  • Skills: error handling patterns
  • 提取内容：gofmt、golangci-lint
  • 钩子：gofmt自动格式化
  • Skills：错误处理模式

Rust

Rust

  • Extract: rustfmt, clippy
  • Hooks: rustfmt auto-format
  • Skills: error handling, async patterns
  • 提取内容：rustfmt、clippy
  • 钩子：rustfmt自动格式化
  • Skills：错误处理、异步模式