agentsclimarketplace

Xiaohongshu to markdown

Skill kohoj/skills/xiaohongshu-to-markdown

Xiaohongshu Article to Markdown — Extracts text from Xiaohongshu image-based posts via OCR. Use when the user needs to convert a Xiaohongshu/Little Red Book post into readable Markdown text. The tool downloads all carousel images from a post and uses macOS Vision OCR to extract the text content. Requires a Chromium browser logged into xiaohongshu.com with CDP enabled. macOS only (uses Vision Framework for OCR).From its SKILL.md

Install
npx -y skills add kohoj/skills --skill xiaohongshu-to-markdown

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 7 commands, including `pip install playwright httpx` and 6 more.

SKILL.md

2.7 KB, 620 tokens by cl100k_base, as published. Nobody here has run it

Xiaohongshu Article to Markdown

Extracts authentication from a Chromium browser via CDP, fetches post data via httpx, downloads carousel images, and OCRs them into a clean Markdown file using macOS Vision Framework.

Quick Start

python3 <skill-path>/scripts/xhs_to_markdown.py <xiaohongshu_url> [--output-dir DIR] [--cdp-port PORT]

Arguments:

  • url (required): Xiaohongshu post URL (e.g., https://www.xiaohongshu.com/explore/...)
  • --output-dir: Output directory, defaults to current working directory
  • --cdp-port: CDP debugging port, default 9222

Output file: {title}_{noteId}.md

Prerequisites

  1. macOS (uses Vision Framework for OCR — no Python OCR dependencies needed)
  2. Chromium-based browser (Chrome / Edge / Brave / Arc) installed and logged in to xiaohongshu.com
  3. Python dependencies: pip install playwright httpx && playwright install chromium
  4. Swift (pre-installed on macOS)

Launch browser with CDP:

# Arc
/Applications/Arc.app/Contents/MacOS/Arc --remote-debugging-port=9222
# Chrome
/Applications/Google\ Chrome.app/Contents/MacOS/Google\ Chrome --remote-debugging-port=9222
# Edge
/Applications/Microsoft\ Edge.app/Contents/MacOS/Microsoft\ Edge --remote-debugging-port=9222
# Brave
/Applications/Brave\ Browser.app/Contents/MacOS/Brave\ Browser --remote-debugging-port=9222

Architecture

URL → CDP(extract cookies) → httpx(fetch HTML) → parse __INITIAL_STATE__ JSON
→ extract HD image URLs → httpx download images → Swift Vision OCR → Markdown
StepMethodNotes
Cookie extractionCDP ctx.cookies()Invisible, no tabs opened
Page fetchhttpx GETWith browser cookies for auth
Data parsing__INITIAL_STATE__ JSONEmbedded in page HTML
Image downloadhttpx concurrentStrips quality compression suffixes
OCRmacOS Vision Framework (Swift).accurate level, zh-Hans + zh-Hant + en
OutputMarkdown fileTitle + meta + OCR text + description

Output Format

# {title}

> Author: {author} | {time} | [Original]({url})

{OCR text from image 1}

{OCR text from image 2}

...

---

{post description, if present}

Lessons Learned

DateLessonAction

What ships with it: 2 files

22.1 KB alongside SKILL.md, 2 of them executable

scripts/

Keep looking

Skills are one crate of 611,884. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.