StripHTML: Save AI Tokens

Raw page HTML is bloated with SVG icons, inline styles, tracking tags, and endless wrapper <div>s. Paste that into ChatGPT, Claude, or Gemini and you burn tokens — and money — on noise the model never needed. StripHTML extracts only what the AI actually needs, right from your browser. Click the toolbar icon on any page, pick a format, and copy. THREE OUTPUTS • Markdown — Clean, readable content with embedded structured data (JSON-LD, __NEXT_DATA__) preserved. Use it when you want the page's INFORMATION as context. • Clean HTML — The page's DOM with structure intact but noise stripped: only interaction-relevant attributes (id, role, aria-*, etc.) remain. Use it when the AI needs to know WHERE and WHAT to click. • Interactive Map — A compact list of every clickable and fillable element, each with a robust CSS selector. The most token-efficient "what can I click here?" format for browser-automation prompts. SEE WHAT YOU SAVE Every output shows the tokens and dollars saved versus pasting raw HTML. Real numbers on a 692 KB Wikipedia article: ~70% saved with Markdown, ~64% with the Interactive Map. On SVG-heavy app and marketing pages, savings reach 80–95%. PRIVATE BY DESIGN Everything runs locally in your browser. No accounts, no servers, no tracking. The page is read on demand only when you click — nothing is sent anywhere. Token counts use an offline heuristic; no network calls. HOW IT WORKS 1. Open any page. 2. Click the StripHTML toolbar icon. 3. Pick Markdown, Clean HTML, or Interactive Map. 4. Hit Copy and paste into your AI. NOTES Restricted pages (browser settings pages, the extension store, raw PDFs) can't be read. If a tab was open before you installed StripHTML, reload it once.