RedNote (Xiaohongshu) Scraper — Trending, Category Feeds & Notes

Pricing: $0.02 per post returned plus a $0.05 Actor Start fee per run. The default limit of 25 posts costs $0.50 + $0.05 = $0.55 per run; $1.00 buys one run of ~47 posts ($0.05 start + 47 × $0.02). Runs that return 0 posts (blocked, failed, or genuinely empty) pay only the start fee.

Pull RedNote (Xiaohongshu, 小红书) trending notes, category discovery feeds, keyword / hashtag / user search results and full note detail as clean JSON. Western brand and sourcing teams use it to read China's earliest publicly observable consumer-demand signal: on RedNote, nearly every viral note is a product review or buying guide, surfacing weeks before Tmall or JD sales rankings. Trending / category feeds and note_detail run without a login; keyword search, hashtag and user-posts modes are login-gated by RedNote and require a logged-in cookie. No Mandarin-speaking analyst in Shanghai required.

中文说明

小红书(RedNote)采集器让你获取小红书的热门内容、分类信息流与笔记数据。每条结果为一篇笔记,可导出为 JSON、CSV 或 Excel,适合品牌、内容营销与消费者研究。

定价:按笔记计费,每条 $0.02,另加每次运行 $0.05 启动费(默认 25 条约 $0.55 一次)。

定时运行:在 Apify 的「Schedules」中新建计划任务,设定关键词或分类,即可定期追踪话题与品类的变化。

AI 智能体:采用按事件计费(PPE),可通过 Apify 的 x402 让智能体用 USDC 直接调用,无需账号。

说明:关键词搜索、话题搜索与用户主页模式必须提供登录 Cookie,否则小红书会弹出登录墙,运行将失败且不产生结果费用;所有字段均来自公开数据。

Modes

Mode	What it does	Cookie?
trending	Homepage discovery feed (category: recommend) or a category channel tab (fashion / food / cosmetics / travel / gaming / fitness / movie / home / love / career / video)	Not needed
note_detail	Full body text + engagement counts (likes / comments / shares / collected) for a note URL or ID. Comment text is not returned — only comments_count.	Not needed
keyword_search	Search by keyword (Chinese or English terms), up to 10 keywords per run	Required
hashtag	Posts for a hashtag / topic (resolved through RedNote's search API), up to 10 tags per run	Required
user_posts	Recent posts by a specific user, up to 10 users per run	Required

topic_probe also exists in the mode list for compatibility. It is a diagnostic mode: it writes to the key-value store only, pushes 0 dataset rows and incurs 0 result charges. It is not a buyer feature.

Cookie requirement, plainly: RedNote serves a login wall to anonymous visitors on search, hashtag and user-profile pages. Live checks on 2026-09-03 without a cookie were blocked in every one of these three modes. Without a valid logged-in cookie those runs fail as BLOCKED with 0 rows and 0 result charges (the start fee still applies). They do not fall back to the explore feed.

Category feeds, plainly: if the requested category tab cannot be activated on the page, the run fails with category tab unavailable (0 rows, 0 result charges). It never delivers the mixed recommend feed labelled as your category. The category field on each record is the note's own text-classified category (beauty / fashion / food / travel / wellness / tech / finance / lifestyle); the channel you asked for is recorded in query.

Time budget: each keyword / hashtag / user is a separate 60–115 s browser session. The run stops starting new queries 45 s before its timeout and delivers whatever the completed queries produced in one push; the run status message then says delivered X records from Q of N queries; N-Q not processed (time budget). Raise the run timeout or split queries across runs for large lists.

Use cases

1. Track emerging Chinese consumer-product trends before they reach Western markets.
2. Brand monitoring via trending and category feeds.
3. Cross-border e-commerce / DTC product research — spot rising categories early.
4. Build a Chinese-language lifestyle-content corpus for NLP / AI.
5. Competitor and KOL discovery via user_posts (with cookie) and note_detail.

Sample input

{
  "mode": "trending",
  "category": "cosmetics",
  "limit": 50,
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["RESIDENTIAL"], "apifyProxyCountry": "CN" }
}

Output fields

Every mode emits the same field set per note:

post_id, author, author_id, title, body, image_urls, hashtags, likes_count, comments_count, shares_count, collected_count, engagement_total (sum of the four counts), posted_at, url, category, mode, query, data_source, as_of_timestamp. note_detail records additionally carry source_url (the URL actually fetched).

Note: the trending / category feed API returns card summaries, so in trending mode body and posted_at can be null. Use note_detail on a record's url to get the full body and timestamp.

{ "post_id": "65a1f...e3", "author": "美妆日记Amy", "title": "早C晚A空瓶记录", "body": null, "likes_count": 12840, "comments_count": 612, "shares_count": 88, "collected_count": 2310, "engagement_total": 15850, "hashtags": ["护肤","空瓶"], "posted_at": null, "url": "https://www.xiaohongshu.com/explore/65a1f...e3?xsec_token=...", "category": "beauty", "mode": "trending", "query": "cosmetics" }

FAQ

Does this RedNote scraper need a login or cookie? trending (all categories) and note_detail do not. keyword_search, hashtag and user_posts do — RedNote shows a login wall to anonymous visitors in those modes and the run fails as BLOCKED (0 rows, 0 result charges) without a valid cookie. Log in at xiaohongshu.com, copy document.cookie, and paste it into the cookie input. Using your own account may violate XHS Terms — use at your own risk.

Can it do keyword search on Xiaohongshu? Yes, with a logged-in cookie. Without one it cannot; it does not substitute the explore feed for a search.

Does it return comments? No. note_detail returns comments_count only; comment text is not extracted.

Is scraping Xiaohongshu / RedNote legal? It pulls only content visible to a logged-in or logged-out visitor of the site as configured; it does not bypass login walls or private profiles. Buyers are responsible for compliance with the site's Terms of Service and applicable law.

What data fields are returned per note? See Output fields above — likes_count, comments_count, shares_count, collected_count, engagement_total, hashtags, posted_at, url and the rest.

What is the maximum limit? 200. The trending / category feed can be scrolled about 8 pages (~230 notes) in one run, and search modes share one time budget, so larger values are not reachable and are not accepted by the input schema.

Do I need to read Chinese? No — fields are structured and English-labeled. Chinese keywords give the best native results; English terms work too.

Related NexGenData actors

• Bilibili Video Search
• China Trends Tracker
• Weibo Hot Search Tracker
• Bilibili Video Comments

---

Built and maintained by NexGenData. Home: thenextgennexus.com.