A Racket job talks to Trends API with one POST to https://api.trendsapi.ai/api, a Bearer key, and a JSON body that names a mode. There is no official Racket library. http-sendrecv ships in net/http-client. read-json ships in json. The first decode reads the envelope, then a second decode reads the body string. Pulled on 2026-11-10, Google Search for racket programming scored 20.0 for the week of 2026-11-07, down 62.26% from 53.0. racket language scored 9.0, up 50.0% from 6.0. Those two tokens closed 11 points apart and with opposite 7D signs. 14D disagreed again: programming printed versus 0.0, language printed 9.0 versus 13.0 (-30.77%). YouTube for racket programming scored 11.0 versus 14.0. Do not copy bare racket Search 18.0 onto a language-adoption alert that only watches the phrase token at 20.0. The same helper that stores those rows also pulls a live Google Trends board. A rank-10 name whose weekly Search still printed 8.0 while daily Search printed 100 is documented in Abdul El-Sayed live rank versus weekly Search.
How does a Racket client call Trends API?
POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json
{"mode":"get_growth","source":"google search","keyword":"racket programming","percent_growth":["7D","30D","3M","12M"]}
#lang racket
(require net/http-client json)
(define (trends-growth key)
(define payload
(jsexpr->bytes
(hash 'mode "get_growth"
'source "google search"
'keyword "racket programming"
'percent_growth '("7D" "30D" "3M" "12M"))))
(define-values (status headers in)
(http-sendrecv
"api.trendsapi.ai"
"/api"
#:ssl? #t
#:method #"POST"
#:headers (list (string-append "Authorization: Bearer " key)
"Content-Type: application/json")
#:data payload))
(define envelope (read-json in))
(read-json (open-input-string (hash-ref envelope 'body))))
http-sendrecv owns the HTTP client. The #:headers list sets the Bearer key. jsexpr->bytes holds the JSON object. The first read-json decodes the transport object into statusCode and body. A second read-json on (open-input-string (hash-ref envelope 'body)) is required before reading recent_value. The collections ship with Racket:
https://docs.racket-lang.org/net/http-client.html
https://docs.racket-lang.org/json/index.html
https://racket-lang.org
http-sendrecv and read-json are client libraries. They are not Trends API sources. There is no Racket package-registry source on this API. raco pkg talks to pkgs.racket-lang.org. That registry is not a Trends source.
The response envelope is {"statusCode": int, "body": string}. body is a JSON string and must be parsed a second time before reading recent_value. An Elm client that does the same double parse with Http.request is on Elm Trends API. A Haskell client is on Haskell Trends API.
What did get_growth return for Racket language tokens?
get_growth for racket programming on google search tips on 2026-11-07. Store the exact strings. racket language is a different series. The two phrases closed 11 points apart. 7D signs disagreed. 14D split them again.
| Keyword | Period | Recent date | Recent | Baseline date | Baseline | Growth |
|---|---|---|---|---|---|---|
| racket programming | 7D | 2026-11-07 | 20.0 | 2026-10-31 | 53.0 | -62.26% |
| racket programming | 14D | 2026-11-07 | 20.0 | 2026-10-24 | 0.0 | vs 0.0 |
| racket programming | 30D | 2026-11-07 | 20.0 | 2026-10-10 | 25.0 | -20.0% |
| racket programming | 3M | 2026-11-07 | 20.0 | 2026-08-08 | 25.0 | -20.0% |
| racket programming | 12M | 2026-11-07 | 20.0 | 2025-11-08 | 73.0 | -72.6% |
| racket programming | YTD | 2026-11-07 | 20.0 | 2026-01-03 | 49.0 | -59.18% |
| racket language | 7D | 2026-11-07 | 9.0 | 2026-10-31 | 6.0 | +50.0% |
| racket language | 14D | 2026-11-07 | 9.0 | 2026-10-24 | 13.0 | -30.77% |
| racket language | 30D | 2026-11-07 | 9.0 | 2026-10-10 | 12.0 | -25.0% |
| racket language | 3M | 2026-11-07 | 9.0 | 2026-08-08 | 24.0 | -62.5% |
| racket language | 12M | 2026-11-07 | 9.0 | 2025-11-08 | 63.0 | -85.71% |
| racket language | YTD | 2026-11-07 | 9.0 | 2026-01-03 | 29.0 | -68.97% |
| racket-lang | 7D | 2026-11-07 | 0.0 | 2026-10-31 | 0.0 | 0.0% |
| racket | 7D | 2026-11-07 | 18.0 | 2026-10-31 | 17.0 | +5.88% |
| racket | 12M | 2026-11-07 | 18.0 | 2025-11-08 | 23.0 | -21.74% |
| racket | YTD | 2026-11-07 | 18.0 | 2026-01-03 | 22.0 | -18.18% |
get_time_series for racket programming on Google Search closed at 20.0 on 2026-11-07, matching get_growth 20.0. Adjacent weeks: 53.0 on 2026-10-31, 0.0 on 2026-10-24, 0.0 on 2026-10-17, 25.0 on 2026-10-10. Peak in this pull is 100.0 on 2026-05-16. Next highest points are 99.0 on 2026-06-06, 98.0 on 2026-04-11 and 2026-04-18, and 97.0 on 2026-05-09. Tip 20.0 is not that May 2026 high. 12M printed 20.0 versus 73.0 (-72.6%), the drop after that spring peak. The payload printed growth 999999 on 14D because that baseline was 0.0. Report 20.0 versus 0.0.
racket language closed at 9.0, up 50.0% from 6.0. Opposite 7D sign from racket programming at 20.0 versus 53.0 (-62.26%). 14D was 9.0 versus 13.0 (-30.77%) while programming 14D printed versus 0.0. Matching Saturday tips are not matching series. Do not copy programming 20.0 onto language 7D +50.0%.
racket-lang printed 0.0 versus 0.0 on every window, with data_quality low. That token is not package-download volume. There is no Racket package-registry source on this API.
Bare racket closed at 18.0, up 5.88% from 17.0. 14D was 18.0 versus 17.0 (+5.88%). 30D was 18.0 versus 18.0 (0.0%). 3M was 18.0 versus 21.0 (-14.29%). 12M was 18.0 versus 23.0 (-21.74%). YTD was 18.0 versus 22.0 (-18.18%). That six-character string mixes tennis equipment and other uses of the word. Do not treat Search 18.0 as Racket language adoption.
Neighbor languages on the same pull, same source, same 7D window, same 2026-11-07 tip:
| Keyword | 7D recent | 7D baseline | 7D growth | 12M recent | 12M baseline | 12M growth |
|---|---|---|---|---|---|---|
| racket programming | 20.0 | 53.0 | -62.26% | 20.0 | 73.0 | -72.6% |
| racket language | 9.0 | 6.0 | +50.0% | 9.0 | 63.0 | -85.71% |
| lisp programming | 32.0 | 24.0 | +33.33% | 32.0 | 65.0 | -50.77% |
| haskell programming | 30.0 | 25.0 | +20.0% | 30.0 | 64.0 | -53.12% |
| ocaml programming | 27.0 | 21.0 | +28.57% | 27.0 | 68.0 | -60.29% |
| clojure programming | 17.0 | 13.0 | +30.77% | 17.0 | 0.0 | vs 0.0 |
| common lisp | 12.0 | 14.0 | -14.29% | 12.0 | 25.0 | -52.0% |
| scheme programming | 7.0 | 4.0 | +75.0% | 7.0 | 32.0 | -78.12% |
| elm programming | 6.0 | 5.0 | +20.0% | 6.0 | 27.0 | -77.78% |
racket programming 20.0 sat 12 points below lisp programming 32.0 and 10 points below haskell programming 30.0. It sat 13 points above racket language 9.0. 7D growth for racket programming printed -62.26% while lisp programming printed +33.33% and scheme programming printed +75.0%. Do not copy lisp 32.0 onto racket programming 20.0. Do not copy scheme +75.0% onto racket language +50.0%. clojure programming 12M printed versus 0.0. Store the zero baseline. High-volume php in this session tipped on 2026-11-07 at 25.0 versus 38.0 with estimated volume 556,000. That is the same Saturday tip as the language keywords, not the older 2026-10-31 tip that still held people-keywords. Store the session pull and the exact recent_date per keyword. A Clojure Trends API helper stores the Lisp-family neighbor on the same 7D window. A Gleam Trends API helper stores gleam programming.
What did YouTube, News, and Wikipedia return?
YouTube for racket programming printed 11.0 versus 14.0 on 7D. Keyword Google News for both Racket phrases printed zeros. Wikipedia titles resolve on their own. Compare Wikipedia on volume.
YouTube for racket programming tipped 11.0 on 2026-11-07 versus 14.0 (-21.43%), matching get_time_series 11.0, while Search sat at 20.0. The payload marked data_quality low. Store the 11.0. YouTube 11.0 is not Search 20.0. Peak in the YouTube series is 100.0 on 2023-06-03, then 94.0 on 2023-08-12, then 89.0 on 2024-01-06. On 2026-05-16 Search printed 100.0 while YouTube printed 0.0. Adjacent YouTube weeks: 14.0 on 2026-10-31, 0.0 on 2026-10-24. 14D, 30D, 3M, 12M, and YTD all printed versus 0.0 (growth 999999). Report 11.0 versus 0.0. Do not treat the YouTube floor as language adoption while Search sits at 20.0 after a 100.0 close in May 2026.
YouTube for racket language closed at 6.0 versus 11.0 on 7D (-45.45%), also data_quality low. Report 6.0 versus 11.0. Language YouTube 6.0 is not programming YouTube 11.0. Do not copy language YouTube 6.0 onto programming Search 20.0. Longer windows for language YouTube also printed versus 0.0.
Bare racket YouTube closed at 52.0 versus 52.0 (0.0%). 12M was 52.0 versus 60.0 (-13.33%). That is not the phrase series. Do not copy bare YouTube 52.0 onto programming YouTube 11.0.
Keyword Google News for racket programming printed 0.0 versus 0.0 on every window, with data_quality low. Store the zeros. Keyword Google News for racket language printed the same zeros, also data_quality low. Do not fill News from Search 20.0. Keyword News and the Google News Top News board are different feeds. Mode docs sit at https://trendsapi.ai/docs.
Wikipedia for parenthetical title Racket (programming language) resolved. October 2026 scored 10.3 with 3,527 pageviews versus September 7.7 with 3,387 (volume +4.13%). 7D and 14D collapsed (collapsed_range: recent and baseline resolved to the same monthly point). 3M was 10.3 / 3,527 versus July 0.0 / 2,969 (volume +18.79%). The payload printed growth 999999 on 3M because the score baseline was 0.0. Report the volume. 12M was 10.3 / 3,527 versus October 2025 21.8 / 4,147 (volume -14.95%). YTD was 10.3 / 3,527 versus December 2025 23.4 / 4,235 (volume -16.72%). Peak in this pull is 100.0 / 8,384 in September 2022, then 87.0 / 7,682 in November 2022, then 84.6 / 7,552 in October 2022. Short title Racket resolved to a different page. October scored 9.9 with 382 pageviews versus September 0.0 with 343 (volume +11.37%). That language-title page is about 9.2 times the short-title volume. Do not join 382 views onto Search 20.0. Do not treat short-title Racket at 382 views as the language page. The language title is the finding. That is the same direction as Elm and Vala, where the language title was the page to store. Wikipedia source docs sit at https://trendsapi.ai/trends/wikipedia-trends.
There is no Racket package-registry source on this API. npm covers npmjs. python covers PyPI. A Racket watchlist that wants package downloads has no third source to join onto Search 20.0.
How does a watchlist store live boards next to weekly scores?
get_top_trends has no keyword. Pair the live Google Trends board with get_growth on the exact board string. Live rank, 1-day rank_change, and 7-day rank_change are three feeds. The rank snapshot can stamp at the same second as the mover snapshot, or days earlier than Wikipedia Trending.
POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json
{"mode":"get_top_trends","type":"Google Trends","limit":25}
On 2026-11-04T07:01:36Z the live Google Trends board ranked abdul el-sayed at 10, with no row on the 1-day mover list. Wikipedia Trending, YouTube Trending, and Google News Top News stamped 2026-11-10T09:01:35Z, six days later. Weekly Search for abdul el-sayed scored 8.0 versus 7.0 on the 2026-10-31 tip. Daily Search printed 100 on 2026-11-04, then 4 on 2026-11-09. YouTube scored 100.0 versus 25.0 on the 2026-11-07 tip. Wikipedia Trending ranked Abdul El-Sayed at 6 on the current stamp. The helper that stores those rows is the blog in this batch: Abdul El-Sayed live rank versus weekly Search. Do not copy that daily 100 onto racket programming 20.0. Board-shape notes for Google Trends live feeds sit at https://trendsapi.ai/trends/google-trends.
Free tier is 100 successful 200s per month. Only 200 responses count against quota. Five language tokens plus one live board is 6 calls. A Monday run that also pulls YouTube and Wikipedia for the programming phrase is 8 calls. Four weeks is 32 calls and still fits the free cap. Starter is 5,000, Pro 25,000, Business 100,000.
Envelope checklist
POSThttps://api.trendsapi.ai/apiwithAuthorization: Bearer <api_key>.- Parse the transport JSON, then parse the
bodystring a second time. - Store
racket,racket programming,racket language, andracket-langas separate series. - Compare Wikipedia
Racket (programming language)on volume at 3,527 October views, not short-titleRacketat 382. - Pair
get_top_trendsrank with bothrank_changewindows andget_growthon the exact board string. Store each keyword'srecent_date.