REVU Developers#
REVU is the Feedback and Behavior Intelligence platform. Feedback turns what users say, such as app store reviews, into structured intelligence. Behavior captures what users do: pageviews, clicks, sessions, vitals and crawler visits.
These docs cover building on both. The SDKs capture behavior on the web, your server, Android and iOS. They are small, fast, and zero-dependency by design: they capture interactions, identity, and context, then hand off to the server, which does the intelligence. The API takes those events in, and reads Feedback and Behavior insights back out for your scripts and AI assistants.
Pick your platform#
- Web SDK -
@revu-ai/core. Vanilla JavaScript, zero runtime dependencies, about 10 KB. One line to autocapture pageviews and clicks. Stable. - Server SDK -
@revu-ai/server. Reports the crawlers that reach your own web server, including the ones that never run JavaScript. Node, Bun, Deno, Cloudflare Workers and Next.js. Stable. - Android SDK - Kotlin capture core. In development.
- iOS SDK - Swift capture core. In development.
- API - the ingest contract every SDK targets, plus read keys and the MCP server for AI assistants.
Start in 60 seconds (web)#
<script async src="https://cdn.revu.ai/behavior"></script>
<script>
window.revu = window.revu || new Proxy({ q: [] }, {
get: (t, m) => (m in t ? t[m] : (...a) => t.q.push([m, ...a])),
});
revu.init({ apiKey: "revu_pk_..." });
</script>
That is the whole integration. Autocapture starts, identity is assigned, and events batch to ingest. From here, read Concepts for the mental model or the API reference for the full surface.
Built for machines too#
These docs are designed to be read by LLMs and AI agents as much as by people:
- /llms.txt - a curated map of every page, the llmstxt.org convention.
- /llms-full.txt - the entire documentation set in one file for full-context ingestion.
- Raw Markdown - append
.mdto any page URL (for example /web/install.md) to fetch the source without HTML.
Principles#
- Capture only. The SDK never computes metrics; that is the server's job.
- Never crash the host page. Public entry points are wrapped; internal errors are swallowed.
- Redact at source. The SDK captures interactions, never input values.
- Lightweight by design. Zero runtime dependencies, always.