Fuzzy, keyboard-first search for static sites. No backend, no build step, no account — drop in one script and it works on this page. But the real game changer is the DOM extraction layer Reef doesn't just search for things, it can act on them. A revolutionary yet simple tool for AI agents to act on websites, and more.
NPM and jsdelivr build coming soon!
Paste it, done. No bundler, no config file, no signup.
Opens instantly. Arrow keys move, Enter selects, Esc closes.
Indexed and searched entirely in the visitor's browser.
Reef reads your sitemap or the links already on the page the moment it loads.
Content is tokenized and scored entirely client-side — nothing is sent anywhere.
⌘K opens instantly. Fuzzy matching ranks results as the visitor types.
Reef is the only option that indexes live DOM content — headings, buttons, fields, files, media, and structured data — straight from the page. Toggle the tools from our parity plan to see how it stacks up.
"Planned" marks capabilities on Reef's roadmap drawn from our parity plan. Competitor rows reflect each project's current public feature set.
<script src="reef.min.js"></script>
Yes. It's a plain script tag, so it works the same regardless of framework.
No. Indexing and searching both happen in the visitor's browser.
Nothing — queries are never transmitted or logged anywhere.
About 44kb, so it won't noticeably affect page load.