Analysis updated 2026-05-18
Let an AI assistant read and summarize a Xiaohongshu post from a shared link.
Pull a post's title, text, author, and engagement stats without logging in.
Fetch a post's images alongside its text, or skip images to save tokens.
Integrate Xiaohongshu post reading into an MCP toolset.
| usubamayoi/xhs-read-mcp | aaaddress1/vibe-reading | amirhosseinjpl/jpl-sub-processor | |
|---|---|---|---|
| Stars | 24 | 24 | 24 |
| Language | JavaScript | JavaScript | JavaScript |
| Setup difficulty | easy | moderate | moderate |
| Complexity | 2/5 | 2/5 | 3/5 |
| Audience | developer | researcher | ops devops |
Figures from each repo's GitHub metadata at analysis time.
No login or cookies needed, works by parsing JSON already embedded in the page HTML.
xhs-read-mcp is a small tool that lets Claude read posts directly from Xiaohongshu, a Chinese social platform also known as RED or Little Red Book. You give it a Xiaohongshu link, and it returns the post's title, main text, author, engagement numbers like likes and comments, the first screen of comments, and any images attached to the post. The way it works is straightforward. Xiaohongshu's web pages embed a block of JSON data inside the page's HTML, under a variable called INITIAL_STATE, and all of the post's real content lives inside that JSON. The tool sends a request to fetch the raw HTML of the link, pulls out that embedded JSON block, and then parses out the specific fields it needs, such as the title and comments. Because everything it needs is already sitting in the page's HTML, it does not require logging into an account, does not need cookies, and does not need a full browser automation tool like Puppeteer to work. This is built as an MCP server tool, meaning it is meant to be used as part of an MCP toolset rather than run on its own as a standalone app. The README points to a file called xhs_read_snippet.js as the reference implementation to look at. It takes a required url parameter, which can be either a short xhslink.com link or a full xiaohongshu.com link, and an optional include_images parameter that defaults to true, setting it to false skips images and returns text only, which uses fewer tokens. The README also notes that if you are using Claude Code specifically, you do not need this MCP tool at all, since Claude Code can already read a pasted Xiaohongshu link using its own web fetching ability. The README lists a few limits: video posts only return their text and cover image rather than video content, only the first screen of comments is captured, and if Xiaohongshu changes its page structure the parsing logic inside this tool may need to be updated to keep working.
An MCP tool that fetches a Xiaohongshu post link and returns its title, text, author, engagement stats, comments, and images without needing login or a browser.
Mainly JavaScript. The stack also includes JavaScript, MCP.
Setup difficulty is rated easy, with roughly 5min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.