| name | cdp-page-to-md |
| description | Fetch authenticated or JS-rendered web pages via Chrome CDP (DevTools Protocol) and convert them to clean Markdown files. Use when: (1) user provides a URL that requires authentication (e.g. claude.ai, chatgpt.com, Google Docs, Notion, Confluence, Jira), (2) user wants to save/archive a web page as Markdown, (3) WebFetch fails with 403/401 or returns empty content, (4) the page is a JS-rendered SPA that needs a real browser. Trigger words: 'fetch page', 'grab page', 'save as markdown', 'convert to md', 'pull down this page', 'CDP', 'archive this URL'. |
CDP Page to Markdown
Fetch web pages through a local Chrome browser with CDP, extract content, and convert to structured Markdown.
Workflow
Step 1: Ensure Chrome CDP is running
curl -s http://localhost:9222/json/version
If this fails, start Chrome with CDP:
bash ~/.claude/skills/cdp-page-to-md/references/start_chrome_cdp.sh
Close all Chrome windows first (Cmd+Q), then run the script.
Step 2: Fetch page content
Run the extraction script — it handles CDP connection, navigation, scrolling, and text extraction:
python3 ~/.claude/skills/cdp-page-to-md/scripts/fetch_page.py "<URL>"
Options: --port 9222 (CDP port), --wait 6 (seconds after navigation), --scrolls 3 (lazy-load scroll passes).
Output format:
TITLE: <page title>
URL: <final URL>
LENGTH: <text length>
---CONTENT---
<raw page text>
Set Bash timeout to 60s+.
Step 3: Convert raw text to Markdown
This is the only step requiring intelligence. Take the raw text output and:
- Use
TITLE as # heading
- Add
> Source: <URL> metadata
- For chat pages (claude.ai, chatgpt.com): separate user/assistant turns with
--- and ## User / ## Assistant headers
- For general pages: infer heading hierarchy, lists, code blocks, bold/italic from content structure
- Remove navigation chrome, footers, cookie banners, "Report" links
- Write
.md file to user-specified path
Site-Specific Notes
| Site | User turn marker | Assistant turn marker | Notes |
|---|
| claude.ai/share/* | Raw user text blocks | Prose blocks after date | Title in page title before | Claude |
| chatgpt.com/share/* | "你说：" / "You said:" | "ChatGPT 说：" / "ChatGPT said:" | May include "Reasoned for Xm Ys" |
References
scripts/fetch_page.py — CDP extraction script. Run directly, no need to read into context.references/start_chrome_cdp.sh — Launch Chrome with CDP enabled (port configurable, default 9222).
