Getting started
Introduction
Imitation Engine is a Claude Code plugin that clones production websites into pixel-perfect, fully interactive mirrors for RL training gyms — every route, popup, dropdown and interaction, without a human in the loop.
What is Imitation Engine?
RL training environments need realistic websites — lots of them, with behaviour that matches production down to the pixel. Building those mirrors by hand is slow, and keeping them faithful is slower. Imitation Engine turns the whole job into a single command inside Claude Code:
/imitation-engine:master https://example.comPoint it at a URL (optionally with an issue list from a spreadsheet, Asana, Slack or a plain text dump) and it crawls the site end to end, registers a project on your dashboard, and clones the pages you pick.
How it works
A project is a website; each website has many pages. Every run moves through three stages:
- 01CrawlA Playwright-driven crawler walks the site breadth-first — following links, opening popups and dropdowns, collapsing dynamic routes into {id} templates — and produces a nested page hierarchy.
- 02ReviewThe discovered pages appear as a file tree on your dashboard. You tick the ones worth cloning and press Start cloning — the only human step in the pipeline.
- 03CloneEach selected page is rebuilt into a pixel-perfect mirror with matching behaviour: navigation, dropdowns, modals and interactions all work like the original.
Progress streams back to the dashboard as crawling → cloning → cloned, and re-running the command from the same directory updates the same project.
What you'll need
- Claude Code installed and signed in.
- An Imitation Engine subscription — one subscription covers one developer (one Claude email, one seat).
- Around 500 MB of disk for the crawler runtime and its Chromium build.
Note