Pass one feed type to get_top_trends. Read the default 25-row window, the mixed board when category is blank, and the empty 7d rank_change note on monthly feeds.
Live data as of 2026-09-04
get_top_trends on Trends MCP returns one live ranked feed per call. REST mode is get_top_trends. The alias is top_trends. type is required. A REST body that omitted type on 2026-09-04 returned HTTP 200 with statusCode 400 and error invalid_request, plus the 24 valid type names. Omit limit and the payload prints limit 25, offset 0, and count 25. google trends and GOOGLE TRENDS both echoed type Google Trends. Category-required feeds do not 400 when category is blank; they return a mixed board of rank-1 rows. sort rank_change on Top Websites with window 7d returned count 0 and a note to retry 30d. Product docs bill MCP at 1 credit per 10 requested rows, max 10 credits per call. REST counts 1 request per call.
From the Trends MCP pull of 2026-09-04, a Google Trends call that sent only type printed limit 25, offset 0, count 25, and as_of_ts 2026-09-04T04:01:37.490666+00:00. A follow-up with limit 1 printed count 1 on the same stamp. The tool did not invent a default in the client and hide it; the server echoed 25.
count is the number of rows in data, not a promise that the board is 25 deep. Short feeds still print count below the requested limit. That cap table lives on the get_top_trends pagination page. This page only needs the default: skip limit, read 25.
Row shape on the default sort is a tuple. Google Trends limit 1 returned [[1, "peter cullen"]]. Steam Most Played limit 1 returned [[1, "Counter-Strike 2"]] at as_of_ts 2026-09-04T06:01:45.861783+00:00. GitHub printed fmtlib/fmt. Wikipedia Trending printed Gloria Steinem. Open Library Trending Books printed Atomic Habits by James Clear. Reddit Hot Posts lagged the 06:01 cluster by two hours, as_of_ts 2026-09-04T04:01:37.490666+00:00. Those strings will move. The tuple shape will not.
A REST POST to https://api.trendsmcp.ai/api with {"mode": "get_top_trends", "limit": 1} returned HTTP 200. The envelope statusCode was 400. The parsed body was error invalid_request and a message that started The 'type' parameter is required. then listed 24 names:
Amazon Best Sellers Top Rated, Amazon Best Sellers by Category, App Store Top Free, App Store Top Paid, GitHub, Google News Top News, Google Play, Google Trends, Google Trends by Category, IMDb MOVIEmeter, Open Library Trending Books, Reddit Hot Posts, Reddit World News, Top Websites, Spotify Top Podcasts, Steam Most Played, Substack, Substack by Category, TikTok Trending Hashtags, TikTok Trending Hashtags by Category, TikTok Trending Searches, Wikipedia Trending, X (Twitter) Trending, YouTube Trending.
Product docs still call that failure missing_parameter. This pull printed invalid_request. Read the live error field.
mode top_trends with type Google Trends and limit 1 succeeded and echoed type Google Trends. A misspelled type GoogleTrend returned 400 invalid_request with Unknown type 'GoogleTrend' and the same 24-name list.
Lowercase google trends and uppercase GOOGLE TRENDS both returned 200 and echoed Google Trends. Product docs still say feed labels are case-sensitive. This pull folded case and canonicalized the string. A parser that rejects anything but the exact title case will refuse calls the server accepts.
get_top_trends does not 400 when category is omitted on the five feeds that docs mark as category-required. The 2026-09-04 pull returned mixed rank-1 rows instead.
Amazon Best Sellers by Category, limit 5, no category, as_of_ts 2026-08-31T04:01:36.214924+00:00:
| Rank | Category field |
|---|---|
| 1 | Industrial Scientific |
| 1 | Kitchen Dining |
| 1 | Patio Lawn Garden |
| 1 | Video Games |
| 1 | Health Household |
Google Trends by Category, limit 5, no category, as_of_ts 2026-09-04T06:01:45.861783+00:00: rank 1 from Politics, Shopping, Hobbies and Leisure, Business and Finance, and Entertainment.
Top Websites, limit 3, no category, as_of_ts 2026-09-02T04:01:41.108889+00:00: three rank-1 rows with category_path values arts-and-entertainment/humor, arts-and-entertainment/books-and-literature, and arts-and-entertainment/animation-and-comics.
Substack by Category, limit 5, no category, as_of_ts 2026-09-04T06:01:45.861783+00:00: rank 1 from Music, U.S. Politics, Health Politics, Culture, and News.
The products and newsletters on those rows will rotate. Five rank-1s from five categories is the durable tell that the feed mixed boards. Pin one board with the official name. Official Amazon names for a single-board call are on the Amazon Best Sellers by Category API.
A wrong name is a 400. Substack by Category with category Science returned invalid_request: Unknown category 'Science'. Valid categories: Art & Illustration, Business, Culture, Faith & Spirituality, Fashion & Beauty, Finance, Food & Drink, Health Politics, Music, News, Sports, Technology, U.S. Politics, World Politics. That is 14 names. Product llms.txt still lists 31, including Science. News on the same type returned 200 with "category": "News" in the payload.
Default sort is current rank, and data is an array of tuples. sort rank_change changes the row type.
Google Trends sort rank_change, window 7d, limit 3, as_of_ts 2026-09-04T06:01:45.861783+00:00, baseline_ts 2026-08-28T02:01:40.117111+00:00: each row was an object with rank, keyword, prev_rank, and rank_change. No category_path. Steam Most Played with the same 7d window returned one object: rank 13, keyword Slay the Spire 2, prev_rank 21, rank_change 8, baseline_ts 2026-08-28T02:01:40.117111+00:00. Daily feeds had a baseline about seven days back.
Top Websites did not. sort rank_change, window 7d, limit 5 printed as_of_ts 2026-09-02T04:01:41.108889+00:00, baseline_ts null, count 0, data [], and note: No baseline snapshot within 3 days of 2026-08-26 for 'Top Websites'. This feed updates roughly monthly. Retry with window="30d".
The 30d retry on the same stamp returned count 3. baseline_ts was 2026-08-02T04:01:35.075212+00:00. Rows were objects with an extra category_path. One row was prudentinfo.com at rank 3, prev_rank 48, rank_change 45, category_path health/geriatric-and-aging-care. The domain will move. The empty 7d note plus a filled 30d object row is the cadence rule.
Feed clocks on this pull were not one timestamp. Amazon Best Sellers Top Rated limit 1 shared the Amazon category stamp 2026-08-31T04:01:36.214924+00:00, four days behind the 06:01 Google, Steam, GitHub, Wikipedia, Open Library, and Substack cluster. Top Websites sat on 2026-09-02. Reddit Hot Posts sat on 2026-09-04T04:01:37. Read as_of_ts per type. Do not assume one snapshot time across 24 feeds.
For percent change on a keyword series instead of a live board, use get_growth. For the stored weekly or monthly points behind that percent, use get_time_series. App Store and Google Play boards are live charts. App history still needs source app downloads or app rankings and an Android bundle ID.
MCP tool name is get_top_trends. REST mode is get_top_trends (alias top_trends). One type per call. limit defaults to 25. offset defaults to 0. sort defaults to rank. window is only read with sort rank_change and accepts 1d, 3d, 7d, 14d, or 30d (default 30d).
{
"mode": "get_top_trends",
"type": "Google Trends"
}
{
"mode": "get_top_trends",
"type": "Amazon Best Sellers by Category",
"category": "Toys Games",
"sort": "rank_change",
"window": "7d",
"limit": 10
}
REST envelope is always HTTP 200. Check statusCode, then parse body. The omit-type call in this pull was HTTP 200 / statusCode 400. MCP failures set isError on the tool result.
Product docs bill a successful get_top_trends call as 1 REST request, or on MCP as 1 credit per 10 requested rows with a max of 10 credits, which is a limit of 200. Failed 4xx envelopes do not count. Free is 100 requests per month. Starter is $19 for 1,000. The Google Trends board without these contract details is the Google Trends Trending Now API.
FAQ