3 steps · 鹦鹉AI
Category content intel: what TikTok and Xiaohongshu are saying about a product category
Give your agent a category keyword and get back the top posts on TikTok and Xiaohongshu (RedNote) with engagement, plus the follower counts of the creators behind the TikTok hits. 3 steps through 鹦鹉AI. Low frequency, high decision value.
One prepaid balance · no per-tool signup · no card
Try it
One prompt runs the whole thing
Set your agent up, once
set up data capabilities — https://parrot.amorist.cn/llms.txt
Then ask for it
Using 鹦鹉AI, search TikTok for 'protein powder' sorted by likes and Xiaohongshu for '蛋白粉' sorted by popularity. Show me the price first. Return a CSV with platform, likes, comments, plays, caption and posted date, sorted by likes. Then pull the profiles of the top 10 TikTok creators and add their follower counts.
Search both sources
TikTok shows Western demand, Xiaohongshu shows Chinese consumer behaviour. One prompt, two markets.
Use the native language on Xiaohongshu
Search the Chinese term. An English keyword returns a thin, unrepresentative page.
Sort by engagement
Both routes accept a sort. Likes-first surfaces what resonated, not what was posted last.
Profile the creators behind the hits
Follower count tells you whether a hit came from reach or from resonance.
The steps
What 鹦鹉 AI calls, and what each call costs
Prices come from the catalog's own rate book, read when this page loads, on one prepaid balance.
| # | Step | What the agent asks | Price |
|---|---|---|---|
| 1 | Search TikTokOne page per call. Region defaults to US. | keyword search with sort and publish-time filters, 20 videos per page | 0.7 credits/success |
| 2 | Search XiaohongshuBilled only on success; errors are free. Returns 20 notes per page. | keyword search with popularity sort and a publish-time window | 15.5 credits/success |
| 3 | Profile the top creatorsRuns once per creator you keep. Charged on 2xx only. | follower count and video count for one TikTok handle | 0.7 credits/success |
At the rates above, 10 rows where every call hits comes to 23.20 credits. The receipt below is what it actually cost.
The receipt
What it actually cost, in credits
Run on 2026-09-14, 10 creators in.
- TikTok
- 20 videos for 'protein powder', top post 719,160 likes
- Xiaohongshu
- 20 notes for '蛋白粉' in the last six months, top note 48,588 likes
- Creators profiled
- 10 of 10 TikTok creators, 11,500 to 11,000,000 followers
- Total metered
- 22.40 credits: 0.70 credits TikTok search, 15.40 credits Xiaohongshu search, 7.01 credits profiles
Twelve calls. The two searches returned 40 posts; the ten profile pulls added follower counts to the TikTok hits. The spread is the finding: the same category's top TikTok post has fifteen times the likes of the top Xiaohongshu note, and the creators behind the TikTok hits range from an 11k-follower account to an 11M one, so reach alone does not explain the ranking. The Xiaohongshu search is the expensive call; everything else is TikHub's per-success rate.
Questions
Before you start
What does searching both sources cost?
The receipt above is one real run. The Xiaohongshu call is the expensive one; the TikHub calls are the cheap ones.
Can I get transcripts?
Yes. The catalog has TikTok transcript routes; add them as a step for the posts you shortlist.
What about Instagram or YouTube?
Add those searches to the prompt. The catalog has keyword search on both.
How do I track this over time?
Run the same prompt weekly and keep the CSVs. The agent can diff the top posts between runs.
Why did my DIY TikTok scraper die?
Endpoints and anti-bot checks move. That is expected. This workflow uses maintained tools so the maintenance sits with them.
Can one scraper cover both TikTok and Xiaohongshu?
Often not well. Many social scrapers are TikTok-first and lag on RedNote. Treat them as separate steps with separate failure modes.