How to reverse-engineer a live site with Enrichr, step by step
September 15, 2026
Most first Enrichr sessions start the same way: a URL, no plan, and every tool name looking equally plausible. This is the walkthrough we'd give a teammate on day one — a real run, in the order we'd actually reach for each tool, against a heavily-animated marketing site.
Start wide, not deep
Before picking a specific layer, run extract_site_profile. It's
the cheapest call in the whole tool set and it tells you what's actually worth
looking at closer — whether the site is a static Webflow export, a
React/Next bundle, or something driving its own WebGL canvas. Skipping this step
is the single most common way to waste a capture: running
reverse_engineer_webgl against a site with no WebGL on it returns an
honest, correct, useless "nothing detected."
Pick the layer that matches the question
Enrichr's tools split roughly into three layers, and the fastest sessions use exactly one of them per question instead of reaching for the biggest hammer available:
- Structure —
reverse_engineer_domfor markup and layout,extract_design_systemfor the color/type/ spacing tokens a site is actually built on. - Motion —
reverse_engineer_animationsfor GSAP, ScrollTrigger, Framer Motion, and CSS keyframes, with exact timings and easing curves, not approximations. - 3D —
reverse_engineer_webglfor shaders, draw calls, and scene graphs when a canvas is doing real work.
Each of those holds a real browser open for seconds to minutes and costs credits, so the rule we follow ourselves: the narrowest tool that answers the question, not the broadest one that might.
> reverse_engineer_animations({ url: "https://example.com" })
{
"library": "GSAP 3.13 + ScrollTrigger",
"timelines": 4,
"scrollTriggers": [
{ "trigger": ".hero", "scrub": 1.2, "start": "top top", "end": "+=120%" }
],
"easing": ["power3.out", "cubic-bezier(.16,1,.3,1)"]
}
reverse_engineer_animations call is enough to know a hero uses a
1.2-second scrub tied to scroll, not autoplay — no need to spend a full
capture to learn that.
When you actually want to build from it: rebuild_brief
rebuild_brief is the most expensive call in the set and the only
one written to be built from directly. It returns a brief plus a download URL for
a zip: the rendered DOM, every asset pulled local and rewritten to relative
paths, the CSS that actually applied after JavaScript ran (not the source
stylesheet), screenshots, and the full analysis — opening offline with no
network calls. Reach for it once, after the narrow tools have already told you
what's worth capturing, not as the first call of a session.
What it won't do
Enrichr reports measurements, not opinions, and it reports them for sites you
have a legitimate reason to analyze — your own site, a client's, a
competitor's public marketing page for a pitch. It will tell you the exact
easing curve, the hex values, the shader count; it won't tell you what a design
team intended, and it won't launder someone else's paid work into a resume
project. Point it at what you're allowed to be looking at, and start with
extract_site_profile every time.