Page2AI — Webpage to Markdown 제작자: igorsaevets
Turn the page you are reading into clean Markdown for an LLM. Runs in your browser; nothing is sent to the author or any third party. Keeps code blocks with language labels, tables and headings. Uses a site's own Markdown when it publishes one.
Android용 Firefox에서 이 확장 기능을 열려면 QR 코드를 스캔하세요
확장 메타 데이터
정보
Page2AI converts the page in your active tab into Markdown that an LLM can actually read, and puts it on your clipboard. One click, or Alt+Shift+M.
What it is for. Pasting a rendered documentation page into a chat gives the model navigation, cookie banners and broken code fences. Page2AI gives it the article: headings, tables, lists, and code blocks with their language labels intact.
What it does with your data. Nothing leaves your machine for the author or any third party. There is no server, no account, no analytics, no telemetry, no crash reporting and no third-party SDK. The extension asks for four permissions and no host permissions at all: activeTab, scripting, clipboardWrite, storage.
The one request it makes. Many documentation sites publish a machine-readable Markdown copy of each page. When you invoke the extension it looks for one on the site you are already reading, and only that site: /llms.txt, /docs/llms.txt, /llms-full.txt, and the page's own .md file if the index names one. Using the site's own Markdown beats converting rendered HTML. Those are same-origin requests made from the page itself. Set "Official Markdown" to never in the popup to switch it off.
Open source. MIT licensed. Source: https://github.com/igorsaevets/page2ai-extension
Measured, not asserted. Extraction quality is benchmarked against trafilatura, Mozilla Readability, Defuddle and Microsoft MarkItDown on a committed corpus of documentation pages, with the protocol registered before the run: https://github.com/igorsaevets/page2ai-benchmark
What it is for. Pasting a rendered documentation page into a chat gives the model navigation, cookie banners and broken code fences. Page2AI gives it the article: headings, tables, lists, and code blocks with their language labels intact.
What it does with your data. Nothing leaves your machine for the author or any third party. There is no server, no account, no analytics, no telemetry, no crash reporting and no third-party SDK. The extension asks for four permissions and no host permissions at all: activeTab, scripting, clipboardWrite, storage.
The one request it makes. Many documentation sites publish a machine-readable Markdown copy of each page. When you invoke the extension it looks for one on the site you are already reading, and only that site: /llms.txt, /docs/llms.txt, /llms-full.txt, and the page's own .md file if the index names one. Using the site's own Markdown beats converting rendered HTML. Those are same-origin requests made from the page itself. Set "Official Markdown" to never in the popup to switch it off.
Open source. MIT licensed. Source: https://github.com/igorsaevets/page2ai-extension
Measured, not asserted. Extraction quality is benchmarked against trafilatura, Mozilla Readability, Defuddle and Microsoft MarkItDown on a committed corpus of documentation pages, with the protocol registered before the run: https://github.com/igorsaevets/page2ai-benchmark
리뷰어 0명이 0점으로 평가함
권한 및 데이터
추가 정보
- 부가 기능 링크
- 버전
- 1.3.0
- 크기
- 46.12 KB
- 최근 업데이트
- 한 달 전 (2026년 7월 28일)
- 관련 카테고리
- 라이선스
- MIT 라이선스
- 개인정보처리방침
- 이 부가 기능에 대한 개인정보처리방침 읽기
- 버전 목록
- 모음집에 추가