请问HN:开源AI浏览器扩展与保持闭源以实现完整的Chromium分支,哪个更好?
独立创始人正在构建 *Vibe* — 三层 AI 副驾驶系统:
1. *Chrome 扩展*(当前 MVP — API 密钥,页面命令)
2. *Chromium 分支*(完整浏览器,内置 AI 副驾驶)
3. *Playwright 代理*(无头,完整的 DOM + 可访问性树访问)
*关键见解:*
- 扩展 = 有限(没有可访问性树,上下文较弱)
- 浏览器 + Playwright = 全力(HTML 树,实时 DOM,自动化)
我 *原本* 打算将扩展开源(MIT 许可),以获得关注和贡献者。
但现在:*未来 90% 的价值在于浏览器 + 代理*,而不是扩展。
*两难境地:*
- 开源扩展 → 社区增长外壳,但我失去对 *真正* 产品的控制
- 保持完全封闭 → 反馈较慢,验证更困难
*选项:*
1. *源代码可用的扩展*(公开代码,不允许分叉/商业使用)
2. *仅开源扩展 SDK*(用于接入 Vibe 浏览器的 API)
3. *保持完全封闭*,先发布浏览器测试版
LangFuse:MIT 核心 + 收费 EE
Raycast:封闭核心,开放扩展
*HN:*
- 我应该将 *弱* 部分(扩展)开源以建立影响力吗?
- 还是在浏览器 MVP 发布之前保持一切封闭?
- 如何在不泄露护城河的情况下获取早期用户?
旧金山,2 个月的资金储备。
代码库(即将上线):https://github.com/VibeTechnologies/VibeWebAgent
推介:pitch.vibebrowser.app
你会怎么做?
查看原文
Solo founder building *Vibe* — 3-tier AI copilot system:
1. *Chrome extension* (current MVP — API key, page commands)
2. *Chromium fork* (full browser with native AI copilot)
3. *Playwright agent* (headless, full DOM + accessibility tree access)<p>*Key insight:*
- Extension = limited (no accessibility tree, weaker context)
- Browser + Playwright = full power (HTML tree, real-time DOM, automation)<p>I <i>was</i> going to open-source the extension (MIT) to get traction/contributors.
But now: *90% of future value is in the browser + agent*, not the extension.<p>*Dilemma:*
- OSS extension → community grows shell, but I lose control of the <i>real</i> product
- Keep all closed → slower feedback, harder to validate<p>*Options:*
1. *Source-available extension* (public code, no forks/commercial use)
2. *OSS only the extension SDK* (API to plug into Vibe Browser)
3. *Stay fully closed*, launch browser beta first<p>LangFuse: MIT core + paid EE
Raycast: closed core, open extensions<p>*HN:*
- Should I OSS the <i>weak</i> part (extension) to build mindshare?
- Or keep everything closed until browser MVP ships?
- How to get early users without giving away the moat?<p>SF, 2 months runway.
Repo (coming): https://github.com/VibeTechnologies/VibeWebAgent
Pitch: pitch.vibebrowser.app<p>What would you do?