用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/microwind/ai-skills --skill bundle-analyzer命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
基于 SOC 职业分类
正在显示 SKILL.md
| name | bundle-analyzer |
| description | Enhanced skill for bundle-analyzer |
| license | MIT |
详细的分析和验证 bundle-analyzer.
核心原则: Bundle size directly impacts page load time. Smaller bundles = 快速er pages = better UX.
始终:
触发短语:
Problem:
Including entire lodash library for one function
Consequence:
- Bundle 50KB larger
- Slower load
- Slower parse
Solution:
Use cherry-pick imports or smaller libraries
Problem:
All code in single bundle, even admin pages users never visit
Consequence:
- Initial load slow
- Wasted bandwidth
- Slow FCP
Solution:
Route-based code splitting
| 问题 | 解决方案 |
|---|---|
| "Not sure where to start" | 识别 the specific issue using the 常见问题 section above. |
| "性能 is poor" | Profile first to 识别 bottleneck before optimizing. |
| "Not sure about 最佳实践" | 检查 the 验证检查清单 to understand quality standards. |
❌ No 代码 splitting ❌ Large dependencies for small 使用 cases ❌ Duplicate dependencies ❌ No tree-shaking ❌ Development dependencies in production ❌ No compression