npm package-download trends API

npm download interest via the Trends API. Normalized 0-100 history and growth for exact package names.

Key: trendsapi.ai/#get-key. HTTP contract and every source: trendsapi-ai/trendsapi.

Authentication

pip install trendsapi-npm
export TRENDSAPI_KEY=your_key

Python 3.9+. Same key as the HTTP API.

from trendsapi_npm import TrendsAPI

client = TrendsAPI()                    # TRENDSAPI_KEY
# client = TrendsAPI(api_key="YOUR_KEY")

Keyword helpers default to source: "npm". Pass source= to hit any other platform with the same client. Official full client (every source, no preset): trendsapi.

Methods

Method	REST mode	Returns
get_time_series(keyword, source=, data_mode=)	get_time_series	list[TrendsDataPoint]
get_growth(keyword, percent_growth=, source=, data_mode=)	get_growth	GetGrowthResponse
get_top_trends(type=, ...)	get_top_trends	GetTopTrendsResponse

source is lowercase (npm). type is exact (Google Trends). Mixing them is a 400.

from trendsapi_npm import TrendsAPI

client = TrendsAPI()                    # TRENDSAPI_KEY
# client = TrendsAPI(api_key="YOUR_KEY")

series = client.get_time_series("vite")
print(series[-1].date, series[-1].value)

growth = client.get_growth("vite", percent_growth=["3M", "12M"])
print(growth.results[0].growth, growth.results[0].direction)

get_time_series

points = client.get_time_series("vite")

Each point:

Field	Always	Meaning
date	yes	YYYY-MM-DD
value	yes	0-100 index for this series
keyword	yes	Echo
volume	no	Absolute volume when available
source or datatype	no	Pipeline label

Python returns list[TrendsDataPoint]. Use .date and .value, not ["date"]. JS returns the same fields as object properties.

get_growth

g = client.get_growth("vite", percent_growth=["12M", "3M", "YTD"])
print(g.results[0].growth, g.results[0].direction)

percent_growth default: ["12M"]. Presets: 7D 14D 30D 1M 2M 3M 6M 9M 12M/1Y 18M 24M/2Y 36M/3Y 48M 60M/5Y MTD QTD YTD. Custom: {"name": "Launch", "recent": "2024-06-01", "baseline": "2024-01-01"}.

Field	Meaning
search_term	Keyword
data_source	Source
results	One object per window (period, growth, direction, dates, values)
metadata	Counts / success flag

Several windows still count as one request. Python: growth.results[0].growth. JS: growth.results[0].growth.

Async

import asyncio
from trendsapi_npm import AsyncTrendsAPI

async def main():
    c = AsyncTrendsAPI()
    return await asyncio.gather(
        c.get_time_series("vite"),
        c.get_time_series("vite", source="google search"),
    )

asyncio.run(main())

Each 200 is one billed request.

Pandas

from dataclasses import asdict
import pandas as pd
from trendsapi_npm import TrendsAPI

df = pd.DataFrame(asdict(p) for p in TrendsAPI().get_time_series("vite"))
df["date"] = pd.to_datetime(df["date"])
print(df.set_index("date")["value"].resample("ME").mean().tail())

Call (curl)

Field	Value
Endpoint	POST https://api.trendsapi.ai/api
Auth	Authorization: Bearer $TRENDSAPI_KEY
History	source: npm with get_time_series or get_growth
Keyword	Exact package name, case-sensitive: vite, @tanstack/react-query
Live type	n/a

curl -sS -X POST https://api.trendsapi.ai/api \
  -H "Authorization: Bearer $TRENDSAPI_KEY" \
  -H "Content-Type: application/json" \
  -d '{"mode":"get_time_series","source":"npm","keyword":"vite"}'

Source notes

• Do not URL-encode @ or / inside JSON.
• One package per request.
• For exact integer counts use api.npmjs.org/downloads.

Errors

HTTP	Client
200	Parsed payload. Python dataclasses / JS typed objects
400	Raises. Fix source or type spelling
401	Raises. Check TRENDSAPI_KEY
404	Raises. No series for that keyword. Do not retry
429	Raises. Quota
5xx	Client retries, then raises

The HTTP body field is a JSON string. SDKs decode it. Raw curl must parse body a second time.

Site: https://trendsapi.ai/trends/npm-trends.

License

MIT. See LICENSE.