Claude Code 入门指南：从零到一搭建你的 AI 工作台
手把手教你安装、配置 Claude Code，安装第一个智能体，10 分钟上手 AI 编程助手。
Claude Code 入门指南
什么是 Claude Code？
Claude Code 是 Anthropic 推出的终端 AI 编程助手，可以直接在你的命令行中工作。
安装
npm install -g @anthropic-ai/claude-code
第一个智能体
安装完成后，试试这个安全审计智能体：
claude "帮我用 skill-security-scan 扫描我的项目"
核心功能
- 智能体市场：1100+ 预置智能体，覆盖安全、开发、运维、营销
- 安全审计：双引擎扫描，先审后用，确保安全
- 工作流编排：组合多个智能体完成复杂任务
最佳实践
- 先用安全扫描器审查智能体
- 从小任务开始，逐步复杂
- 善用收藏和工作流功能
每天访问 TG960W 查看最新智能体和教程。
相关智能体
secure-claude-code
Harden Claude Code security with defense-in-depth: deny rules, PreToolUse hooks, OS sandbox, and Docker isolation for VPS/remote environments. Use when setting up a new machine, new project, or hardening an existing Claude Code installation. Triggers: security setup, harden claude, secure claude code, sandbox setup, docker sandbox, vps security, destructive command protection
skill-shielder
Security audit tool for Claude Code skills and repositories. Scans for prompt injection, dangerous scripts, supply chain attacks, and permission scope issues before you install or trust a skill. Use when: (1) evaluating a new skill or MCP before installing, (2) auditing a GitHub repo for safety, (3) reviewing downloaded code for hidden threats, (4) checking if a skill is safe to use. Trigger phrases: "audit this skill", "is this safe", "scan this repo", "security check", "skill check", "shield this", "evaluate this skill"