Page2AI — Webpage to Markdown

Convert any webpage to clean, AI-ready Markdown. 100% local, open source.

Page2AI turns any webpage into a clean, AI-ready Markdown document with one click — ready to paste into your LLM, coding agent, or RAG pipeline. Everything runs inside your browser. No servers, no accounts, no telemetry, no cloud. WHAT MAKES IT DIFFERENT • Profile-aware extraction — auto-detects the site kind (docs, marketing, research, dashboard) and tunes the strategy per profile. Manual override in one click. • Tab & dropdown capture — pulls code from hidden tabs (Python vs TypeScript vs cURL vs …) with DOM-position-aware dedup, so you get the code from every tab, not just the active one. • Code blocks with language hints — ```python, ```js, ```bash preserved verbatim so LLMs read them correctly. • MDX / JSX post-processing — Mintlify components (<Note>, <CodeGroup>, <Tabs>, <AccordionGroup>) become plain Markdown. • llms.txt short path — if the site publishes an official .md alongside the page, Page2AI uses it directly. • Quality gate — post-extraction check on <pre> count vs a plain-text baseline; automatic fallback if under-extraction is detected. • Structured-data appendix — JSON-LD, OpenGraph, Microdata, Next.js __NEXT_DATA__ hoisted into a machine-readable appendix at the end. • Cached-result recovery — close the popup mid-run, reopen it, your result is waiting (badge shows ✓). • Dark mode, live progress log, PII masking (opt-in). HANDLES JAVASCRIPT-HEAVY SITES Modern web apps often defeat plain-text readers with client-side rendering, hidden tab content, or code shown one language at a time. The profile system detects the page kind and applies rules that surface content plain readers miss. FOR WHOM • Researchers feeding docs to their AI assistant. • Developers preparing README-ready snippets from API docs. • Writers clipping web articles into their notes app. • Anyone who used MarkDownload-style clipping — this is a modern replacement. PRIVACY Page2AI does not send data anywhere. No analytics, no telemetry, no crash reports, no cloud service, no accounts, no remote code. The extension only reads the page you explicitly click on (via activeTab), converts it to Markdown on your device, and copies it to your clipboard. Everything else is preferences stored locally in chrome.storage. MINIMUM PERMISSIONS • activeTab — read only the page you clicked on. • scripting — inject the reader script into that page. • clipboardWrite — copy the result to your clipboard. • storage — remember your profile preference between sessions. No <all_urls>. No host_permissions. No tabs API. Chrome will not warn you that this extension can read all your data on all sites — because it cannot. ALSO AVAILABLE ON FIREFOX addons.mozilla.org/en-US/firefox/addon/page2ai-webpage-to-markdown/ FOR AI AGENTS (MCP) Optional MCP server lets AI agents call Page2AI directly through the Model Context Protocol. Install with: npm i @page2ai/mcp — source at github.com/igorsaevets/page2ai-mcp. OPEN SOURCE MIT-licensed. Full source at github.com/igorsaevets/page2ai-extension. Verify every claim in this listing against the code. HOTKEY Alt+Shift+M opens the popup; Enter runs the extraction. BUILT BY Igor Saevets, AI Expert and Entrepreneur.