Hi Product Hunt! I'm Igor, the maker of Page2AI. THE PROBLEM, IN ONE STORY. You find the perfect API doc and paste it into Claude or ChatGPT. The code example on that page has three tabs: Python, TypeScript, cURL. Ctrl+A copies only the tab that is open. The other two are hidden in the DOM and never reach your AI. Code blocks arrive without language tags, tables collapse, and the cookie banner rides along. WHAT IT COSTS YOU. Either you clean this up by hand, a few minutes per page, many times a day, or you feed the mess to the model and get worse answers: wrong-language snippets, broken code, made-up table values. On API models the junk also bills you as tokens. WHAT PAGE2AI DOES. One click (or Alt+Shift+M). The page becomes clean Markdown: real headings, fenced code with language tags, every tab of every code group including the hidden ones, proper tables, and a YAML frontmatter block with title, author, date and canonical URL. The result lands on your clipboard. WHO IT IS FOR: - Product managers: competitor pages, changelogs and pricing pages become clean notes your AI can actually compare. - Entrepreneurs: market research without an hour of copy-paste cleanup. - Developers: docs go into Cursor, Claude Code or your RAG pipeline with structure intact. - IT teams: vendor docs and runbooks land in Obsidian or Notion as proper Markdown. PRIVACY BY CONSTRUCTION. Everything runs inside your browser. No servers, no accounts, no telemetry. The extension can only read the tab you click on (activeTab; no all_urls). MIT-licensed, full source on GitHub, and a public reproducible extraction benchmark you can run yourself. There is also an MCP server (npx page2ai-mcp) so Claude and other agents can pull clean pages themselves. Tell me where extraction breaks for you. I read everything and ship fixes fast.

Page2AI

Webpage to clean Markdown for AI, hidden code tabs included

Webpage to clean Markdown for AI, hidden code tabs included