Pass source wikipedia for a monthly page-view series, or type Wikipedia Trending for the live article board. The two strings are not interchangeable.
Live data as of 2026-09-09
A Wikipedia page views vs Wikipedia Trending comparison on Trends MCP is two different tools. Source wikipedia on get_growth or get_time_series returns a monthly 0-100 page-view series, with a volume field when the pipeline has it. Type Wikipedia Trending on get_top_trends returns ranked article titles and takes no keyword. From the September 9, 2026 pull (as_of 2026-09-09T04:01:41 UTC), rank 1 was Elizabeth Holmes, and that title printed wikipedia 30D index 0.8 and volume 167 on recent_date 2026-08-01. Rank 8 Deaths in 2026 printed index 100.0 and volume 4,432,210. 22 of the 25 board labels returned HTTP 404 no_data. 7D and 1W collapse on this monthly grain. Mixing the two strings is a 400.
The September 9, 2026 board is the split that still matters next month: live rank is not a volume ranking of last month's wikipedia series, and most board labels have no keyword series at all. Omitting limit returned limit 25, count 25, offset 0, as_of_ts 2026-09-09T04:01:41.931714+00:00. Rank 1 was Elizabeth Holmes. Rank 2 was Killing of the Clancy children. Follow-up get_growth calls on source wikipedia used recent_date 2026-08-01, one month behind the board stamp:
| Board rank | Article | 30D index | 30D volume | 30D change |
|---|---|---|---|---|
| 1 | Elizabeth Holmes | 0.8 from 18.6 | 167 from 234 | -95.7% |
| 2 | Killing of the Clancy children | no_data | no_data | HTTP 404 |
| 8 | Deaths in 2026 | 100.0 from 97.1 | 4,432,210 from 4,304,209 | +2.99% |
| 10 | Nathan Fielder | no_data | no_data | HTTP 404 |
| 19 | September 11 attacks | 7.2 from 2.7 | 447,092 from 334,838 | +166.67% |
| 20 | Spider-Man: Brand New Day | no_data | no_data | HTTP 404 |
| 24 | Coco Gauff | no_data | no_data | HTTP 404 |
Twenty-two of the 25 board labels returned no_data on source wikipedia, including rank 2 Killing of the Clancy children, rank 3 Alex Michelsen, rank 4 Bernard Boursicot, rank 5 Hanuman Ansh, rank 6 Mirzapur (film), rank 7 You Can See Everything, rank 9 Alice Weidel, rank 10 Nathan Fielder, rank 11 Frances Tiafoe, rank 12 The Falling Man, rank 13 Navier-Stokes equations, rank 14 Alternative for Germany, rank 15 Neatsville, Kentucky, rank 16 The Gentlemen (2024 TV series), rank 17 Toxic (2026 film), rank 18 Lanterns (TV series), rank 20 Spider-Man: Brand New Day, rank 21 Navier-Stokes existence and smoothness, rank 22 Anna's Archive, rank 23 G. D. Naidu, rank 24 Coco Gauff, and rank 25 Coyote vs. Acme. Parenthetical titles missed. So did several plain names. Treat a miss as a miss. Do not invent a volume for the rank 2 token.
Among the three labels that returned a series, Deaths in 2026 at rank 8 held about 26,540 times the 30D volume of Elizabeth Holmes at rank 1, and still sat seven places lower. September 11 attacks at rank 19 printed 447,092 volume and +166.67% on the index. Holmes at rank 1 had fallen 95.7% on 167 volume. The board is a ranked title list for that scan. The series is a monthly index with a separate volume column that closed on 2026-08-01.
sort=rank_change with window=7d and limit=10 still used the 04:01:41 stamp, plus baseline_ts 2026-09-02T02:01:31.305488+00:00. The first mover was Labor Day at rank 37 (prev_rank 195, rank_change 158). September 11 attacks sat at rank 19 (prev_rank 144, rank_change 125). Other movers sat at ranks 51 through 134. Mover ranks routinely sit past the default 25-row window. Labor Day on source wikipedia printed 30D index 15.7 and volume 632.
Two strings, two tools. REST uses mode get_growth or get_top_trends. MCP uses the matching tool names. The keyword series page is Wikipedia page views. The live board page is the Wikipedia Trending API. Window defaults live on get_growth. Pagination and limit are documented with get_top_trends.
{"mode": "get_top_trends", "type": "Wikipedia Trending"}
{"mode": "get_growth", "source": "wikipedia", "keyword": "Deaths in 2026", "percent_growth": ["30D", "1M", "3M", "12M"]}
{"mode": "get_time_series", "source": "wikipedia", "keyword": "Deaths in 2026"}
{"mode": "get_growth", "source": "wikipedia, google search", "keyword": "Deaths in 2026", "percent_growth": ["30D"]}
The MCP equivalents are get_top_trends(type="Wikipedia Trending"), get_growth(keyword="Deaths in 2026", source="wikipedia", percent_growth=["30D", "1M", "3M", "12M"]), and get_time_series(keyword="Deaths in 2026", source="wikipedia"). REST top_trends with type Wikipedia Trending and limit 3 on September 9, 2026 returned as_of_ts 2026-09-09T06:01:35.725723+00:00, with count 3: Elizabeth Holmes, Killing of the Clancy children, Alex Michelsen. That later stamp kept the same top three titles as the omitted-limit MCP pull.
1M matched 30D exactly on Deaths in 2026. 3M printed index 100.0 versus 91.1, growth +9.77%, volume 4,432,210 from 4,039,786. Extra windows still sit on one source-plus-keyword request.
The comma-separated Deaths in 2026 call returned sources_requested ["wikipedia", "google search"], sources_successful 2, and trend_analysis.uptrend_ratio 1/2. Wikipedia 30D was +2.99% with 22 monthly points and recent_date 2026-08-01. Google Search 30D was -42.86% (35.0 to 20.0) on recent_date 2026-09-05 with 261 weekly points and no volume field in that row. Same preset, opposite signs, different grain, different close dates. Read source_results before joining those percentages.
get_time_series for Deaths in 2026 on wikipedia returned 22 monthly rows from 2024-11-01 (index 0.0, volume 0) to 2026-08-01 (index 100.0, volume 4,432,210). Each row carried date, value, volume, keyword, and datatype wikipedia. Index 100.0 landed on 2026-08-01. On 2025-08-01 the index printed 0.0 while volume stayed 44. The 0-100 column is not a rescaled volume.
Elizabeth Holmes shows why a 30D percent can look huge on a tiny series. The handler echoed search_term Elizabeth holmes. Index 18.6 / 234 on 2026-07-01 fell to 0.8 / 167 on 2026-08-01, growth -95.7%. Volume fell 28.63%. The index drop is a 17.8-point move on a small baseline, not a 4-million-view collapse. Holmes time series held 32 monthly points from 2024-01-01 (index 0.8, volume 167) to 2026-08-01. Peak index 100.0 landed on 2024-04-01 with volume 541. On 2026-06-01 the index printed 0.0 while volume stayed 164.
get_top_trends with type wikipedia returned HTTP 400, error invalid_request, message Unknown type 'wikipedia', plus 24 valid feed labels that include Wikipedia Trending. get_growth with source Wikipedia Trending and keyword Elizabeth Holmes returned HTTP 400, error invalid_source, message 'wikipedia trending' is not a valid source., plus 16 valid sources: google search, google images, google news, google shopping, youtube, wikipedia, tiktok, reddit, amazon, news sentiment, news volume, app downloads, app rankings, steam, npm, python, all. The handler lowercased the rejected source. It did not map it to wikipedia.
The inverse fold exists on the feed name. type wikipedia trending on get_top_trends returned HTTP 200 and echoed type Wikipedia Trending, with the same 25 titles as the canonical label and as_of_ts 2026-09-09T06:01:35.725723+00:00. llms.txt still describes feed labels as case-sensitive. The live handler accepted this lowercase Wikipedia string on September 9, 2026. Prefer the canonical Wikipedia Trending spelling in stored configs.
get_time_series on wikipedia, google search returned invalid_source too, with the 16 names and no all. Multi-source strings belong on get_growth only.
REST omitting type is a different 400. POST https://api.trendsmcp.ai/api with mode get_top_trends and limit 5, no type, returned HTTP 200 with envelope statusCode 400, error invalid_request, message The 'type' parameter is required, then the same 24-name list. llms.txt still names that miss missing_parameter. Read statusCode inside the JSON, not only the HTTP line.
country GB with limit 5 on Wikipedia Trending returned HTTP 400, error invalid_request, message that country is only supported for Google Trends, Google Trends by Category, TikTok Trending Hashtags, TikTok Trending Hashtags by Category, X (Twitter) Trending, and YouTube Trending. Wikipedia Trending is not on that list. Omit country on this feed.
Omit percent_growth on Deaths in 2026 and REST returned period 12M, recent_date 2026-08-01, baseline_date 2025-08-01, index 100.0 versus 0.0, growth 999999.0, volume 4,432,210 from 44, volume_growth +10,073,104.55%. The default window is 12M. That default ran. The printed percent is the pipeline's answer to a zero index, not a demand figure to put in a forecast. Quote volume_growth when the index baseline is zero.
7D, 1W, and 14D failed with collapsed_range on Holmes, Deaths in 2026, and September 11 attacks. Recent and baseline resolved to the same monthly point. llms.txt still lists 7D among standard growth windows. That window needs denser points than this source had on this pull. The working set on this grain is 30D, 1M, 3M, 12M, and YTD.
Holmes 12M printed -98.18% (44.0 to 0.8) with volume 167 from 330. Holmes YTD printed -96.61% (23.6 to 0.8) with volume 167 from 253. September 11 attacks 12M printed -30.1% (10.3 to 7.2) with volume 447,092 from 524,981. Those percents describe the August 1 close, not the September 9 board.
get_top_trends MCP pricing is 1 credit per 10 requested rows, max 10 credits per call. REST counts 1 request per call. Free boards still cap at the top 10. The 25-row pulls on this page are the paid default, which is also what the handler returns when limit is omitted on a paid key. The Wikipedia pageviews growth API page is the keyword-window follow-up. This split stays inside one platform: monthly page-view history versus live article rank.
FAQ