A TikTok trends API has three jobs: what is trending now, how one hashtag moved, and whether it is still climbing. One REST endpoint covers all three with three tools. get_top_trends on the TikTok Trending Hashtags board returns the current ranked list: on 2026-11-03 the top five were halloween, usa, nfl, meme and football. get_time_series with source TikTok returns daily hashtag views: #halloween ran 28 points from 2026-09-06 to 2026-10-05, ending at an index of 98.5 with 376,569,007 views. get_growth on the same hashtag returns +421.2% over 30 days on the index and +103.3% on raw views. TikTok history is short, so a 90-day growth window fails with date_out_of_range; plan around that.
What each TikTok question maps to
| Question | Mode | Exact call |
|---|---|---|
| What hashtags are trending right now | get_top_trends |
type = TikTok Trending Hashtags |
| Which hashtags climbed this month | get_top_trends |
same board, sort = rank_change, window = 30d |
| How did one hashtag move over time | get_time_series |
keyword = hashtag name, source = TikTok |
| Is the hashtag still climbing | get_growth |
keyword = hashtag name, source = TikTok, percent_growth = ["30D"] |
Every call is POST https://api.trendsmcp.ai/api with Authorization: Bearer YOUR_KEY and a JSON body whose mode picks the tool. The endpoint always answers HTTP 200, so read statusCode first and parse the body string a second time.
get_top_trends: the daily hashtag board
The board needs no keyword. The response below is the real board as of 2026-11-03T09:01:33Z:
{"mode": "get_top_trends", "type": "TikTok Trending Hashtags", "limit": 10}
The board snapshot below is the real response as of 2026-11-03T09:01:33Z. It opened with halloween, usa, nfl, meme and football. Music, citytastetest, dance, animation and targetaudience completed the top ten. The full ranked list is stored in the piece's data file.
limit and offset page the board; the request above used limit 10. The top trends reference documents the parameters.
For rank climbers, sort=rank_change with a window such as 30d returns objects with prev_rank and rank_change instead of plain rows. The same board on 2026-11-03 with a 30-day window showed christapike climbing 70 ranks (171 to 101), october up 60 (91 to 31), and costume up 52 (183 to 131).
get_time_series: daily views for one hashtag
The keyword is the hashtag name, with or without the #. For halloween on TikTok the call returned 28 daily points from 2026-09-06 to 2026-10-05:
{"mode": "get_time_series", "keyword": "halloween", "source": "TikTok"}
The latest point reads an index of 98.5 with 376,569,007 raw views. The series opens on 2026-09-06 at an index of 18.9 with 185,228,587 views and ends on 2026-10-05 at 98.5 with 376,569,007, so the whole month is one long climb. TikTok is one of the sources that carries raw volume alongside the 0-100 index, so both come back in every point. The time series reference documents the point fields; the TikTok board shows this board live on the site.
get_growth: the 30-day climb in index and raw views
One growth call takes several windows at no extra cost. For halloween on TikTok with percent_growth set to ["30D"]:
- Index: 18.9 on 2026-09-06 to 98.5 on 2026-10-05, growth +421.16%, direction increase.
- Raw views: 185,228,587 to 376,569,007, volume growth +103.3%, direction increase.
The response returns both dates and both values per window, so a script can audit the arithmetic. The request used percent_growth set to ["30D"]. Both the index and the volume arithmetic arrive in the same result object, so one call answers the climb question twice: relative interest and absolute views. The 90-day window on this keyword fails with date_out_of_range: the series only holds 28 days of history, and the error says so plainly instead of guessing.
What TikTok coverage does not include
Four boundaries, verified in the calls above. History is short: TikTok series return about a month of daily points, not five years of weekly. The board updates daily, not live. Top posts for a hashtag come from get_top_posts, a separate tool that needs the Business plan (see /pricing). And the volume field counts views on the hashtag page, not unique viewers; compare volumes across hashtags, not against platform totals.
