Why I built bouncy Every time I needed to scrape a JS-rendered page, the choice was: spin up Playwright (200+ MB per page, ~1s cold start, Node + Chromium) or fall back to curl + regex (fast, breaks on anything dynamic). For most jobs (fetch a page, extract a title, run one tiny snippet of JS) both felt wrong. So I built bouncy. A tiny Rust headless browser: 10-21 MB per page, ~30 ms cold start with V8, ships as a single binary. No Chromium, no Node runtime. Has its own DOM, embedded V8, HTTP client, and HTML extractor, all separately usable as crates if you only need one part. It also speaks the Chrome DevTools Protocol, so Playwright / puppeteer-core can drive it as a drop-in for the cases where you don't need real layout or paint. Plus an MCP server so Claude / Cursor / any MCP client can scrape with it natively. If you've ever wished there was a middle ground between Playwright and curl, this is it.

bouncy

Tiny Rust headless browser for scraping.

Tiny Rust headless browser for scraping.