Jina Reader without the rate limits, running in your tab

What it does

JinaClip is a Chrome MV3 extension that converts the page you’re looking at into LLM-friendly Markdown, JSON, HTML, or plain text, and drops it straight into your clipboard. It’s a local reimplementation of Jina AI’s cloud reader (r.jina.ai) — same output philosophy, zero network calls. Trigger it via the toolbar icon, right-click menu, or a global hotkey.

The interesting bit

The extraction engine doesn’t guess where the article lives. It scores candidate elements (div, main, article) by text density minus link density, penalizing nav bars, buttons, and dashboards as noise. It also flattens Shadow DOM and iframes before conversion — something the README claims the cloud version handles poorly.

Key highlights

• 100% local processing; no cloud calls, no rate limits, no token costs
• Heuristic DOM scoring for content-root discovery instead of naive scraping
• Shadow DOM and dynamic iframe support
• Full-page screenshot capture via scroll-stitched OffscreenCanvas, delivered as a Markdown-wrapped Base64 image
• Vanilla ES6+ codebase with a small service-oriented structure (localReader.js is the core engine)

Caveats

• The feature matrix comparing against the cloud reader is the author’s own marketing; no benchmarks or accuracy comparisons are provided
• The pageshot pipeline relies on scroll-and-stitch, which typically struggles with lazy-loaded or animated content — the README doesn’t say how (or if) that’s handled
• No mention of Firefox or other browsers; this is Chrome MV3 territory

Verdict

If you regularly paste web content into LLMs and dislike rate limits or sending URLs to a third-party service, this is a genuinely useful tool. If you need headless or bulk scraping, it’s the wrong shape — it’s a clipboard tool, not a crawler.