Xiaohongshu (RedNote) scraper: what fetching a post takes in 2026, and an API that returns the text instead
Short answer: a plain request for a Xiaohongshu (小红书, RedNote) post gets a login page or a "view it in the app" wall, not the post. The open-source scrapers that do work drive a browser logged in with your account, and what they return is fields and image URLs — the text inside the images and the speech in videos still has to be extracted. If what you want is the post's content as text, a reading API skips both steps: send the share link, get the caption, the text in every image, a video's transcript and the top comments. No account of yours is involved.
Disclosure: LinkDigest is the author's own service. Every "measured" line below was checked on 2026-11-02 from an ordinary home connection.
What a plain request gets
Measured on public posts, one request each:
| Request | What came back |
|---|---|
| Desktop browser, not logged in, full share link | Redirected to /login |
Phone browser, share link with its xsec_token |
The post's data in the page |
Phone browser, the same address without xsec_token |
/404/sec_…, a page that looks like a deleted post |
| Phone browser, an older share link | 40 visible characters: 「当前内容仅支持在小红书 APP 内查看」 ("only viewable in the app") |
The token rule, with nine variants of one link, is written up in Xiaohongshu note link returns 404?. Repeated requests also start failing on their own: the same full link that opened on the first request was refused a dozen requests later.
What the working scrapers do
The popular open-source route, MediaCrawler, runs Playwright with a browser you have logged into with your own Xiaohongshu account and reads the platform's data as that account. It is the right tool for bulk discovery — keyword search, full comment threads, creator pages — and it is free. The costs are real: you maintain the login, the requests are made as your account, and the output is fields plus media URLs. A post written into twelve image cards comes back as twelve image links; the words in them are not there until you run OCR, and a video's words are not there until you run speech recognition.
What a reading API returns instead
One call per share link, no Xiaohongshu account:
curl -X POST https://linkdigest.dev/api/v1/digest \
-H "Authorization: Bearer ld_live_..." \
-H "Content-Type: application/json" \
-d '{"url": "https://xhslink.com/o/...", "format": "json"}'
| You get | Example from real reads |
|---|---|
caption, title, author, posted_at, stats, tags |
The post as written, with likes and collects at read time |
images[] — each image described, with the text inside it |
A public demo: 16 images, 390 lines of image text |
transcript[] for video posts |
Timed speech segments, plus on_screen[] text from the frames |
comments[] |
The top comments the share page shows (5 of 207 on one post read 2026-11-02) |
key_points[] |
Takeaways, each with its quote checked against the post |
translation (with translate_to) |
The same in English or another language |
A read takes seconds to a minute; a link someone already read is served from cache at no charge; a link that cannot be read (deleted, private, missing its token) returns a clear error and costs nothing. Pricing is per post: 1 credit covers a post with up to 6 images.
Which one to use
- Bulk discovery — every post for a keyword, whole comment threads, follower lists: a scraper such as MediaCrawler, or a raw-data API; see the Just One API comparison.
- The content of the posts you care about, as text an LLM can use, without running a logged-in browser: a reading API. Many pipelines do both — discover with one, read the posts worth reading with the other.
There is also a topic search in front of the reader (search_links over MCP, /api/v1/search over REST), which returns a handful of recent posts per query — enough to find posts to read, not a replacement for bulk crawling.
Getting started
Sign-up gives 10 free credits, no card. The Xiaohongshu platform page has the MCP setup for Claude Code and Cursor; the API docs have every field. Prefer to stay on Apify? The same reader is the Xiaohongshu note to text Actor, billed through Apify.