DEV Community: Rimu Labs

The wappalyzer npm package is deprecated: options for detecting tech stacks in code

Rimu Labs — Sun, 18 Oct 2026 03:25:21 +0000

For years, npm install wappalyzer was how developers detected a website's technologies in code. The package's last release, 7.0.3 from August 2023, is now marked deprecated on npm ("Package no longer supported"), and the open-source GitHub repository is no longer public. Wappalyzer continues as a commercial product. If you need tech detection in your own code, here are the options.

Option 1: Wappalyzer's paid service

The browser extension is still the quickest way to check a single site by hand, and Wappalyzer sells API access. That's a good fit if you want a managed service and the pricing works for your volume.

Option 2: the open-source fingerprints plus your own engine

The community fork webappanalyzer (GPL-3.0) keeps the technology fingerprints up to date and is actively maintained. It's data only, so you write the engine:

Fetch the page.
Match each technology's patterns against the HTML, script URLs, headers, cookies and meta tags.
Apply the implies, requires and excludes rules.
Pull versions out of the regex groups, and add up confidence.

A basic matcher is a weekend project. Getting versions, implied technologies and false positives right takes longer.

Option 3: a hosted API built on those fingerprints

We maintain a Tech Stack Detector on Apify that runs that engine for you, against a list of any size. It also reads public DNS, which a browser extension can't see. MX and SPF records show the email provider and sending services, and TXT verification records reveal connected SaaS tools such as Atlassian, DocuSign or Zoom.

From Node.js (npm install apify-client):

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('rimulabs/tech-stack-detector').call({
    domains: ['basecamp.com', 'allbirds.com'],
    includeDns: true,
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
for (const site of items) console.log(site.domain, '->', site.techStack);

Real output from 18 October 2026:

basecamp.com -> Atlassian Cloud, Bootstrap, Cloudflare, Cloudflare DNS, Google Search Console, MailChimp, Microsoft 365, Open Graph, Priority Hints, Stimulus, Zoom
allbirds.com -> Apple Pay, Cloudflare, Cursor, DocuSign, Google Search Console, Google Tag Manager, HSTS, HTTP/3, Microsoft 365, Miro, Open Graph, PayPal, Priority Hints, Shopify, Swiper

Each site also has a technologies array shaped much like the old package's output, plus foundIn, the evidence behind each detection:

[
  { "name": "Shopify", "version": null, "confidence": 100, "categories": ["Ecommerce"], "foundIn": ["meta", "scripts"] },
  { "name": "PayPal", "version": null, "confidence": 100, "categories": ["Payment processors"], "foundIn": ["html"] },
  { "name": "Microsoft 365", "version": null, "confidence": 100, "categories": ["Webmail", "Email"], "foundIn": ["dns:mx", "dns:txt"] },
  { "name": "DocuSign", "version": null, "confidence": 100, "categories": ["Miscellaneous"], "foundIn": ["dns:txt"] }
]

It also has ready-made columns such as cms, ecommerce, analytics, email, hosting and cdn for spreadsheets. A filter returns only sites using the technologies you name, and a monitor mode reports what was added or removed since the last run.

Differences from the old npm package

No headless browser. The old package could drive Chrome and see tools that load only after JavaScript runs. The Actor reads the page as served, which is faster and cheaper, but some tag-manager-loaded pixels can be missed. DNS checks make up for part of that.
Polite by design. It fetches each home page once, respects robots.txt, and reports sites behind a bot check as blocked rather than trying to get around it.
Nothing to host. There are no Chrome binaries or fingerprint updates to manage. Results come back as JSON, CSV or Excel.

Cost: $3.00 per 1,000 sites checked. Sites that are unreachable, blocked or closed to crawlers are free. A weekly check of 50 competitors costs about $0.60 a month.

The full guide, kept up to date, is at rimulabs.com/guides/wappalyzer-alternative.

Disclosure: Rimu Labs builds and maintains the Actor in this post. It was written by the AI assistant for Rimu Labs and checked against real runs on 18 October 2026. Questions or corrections: support@rimulabs.com.

pytrends is archived: how to migrate your Google Trends code (no more 429s)

Rimu Labs — Sun, 18 Oct 2026 03:23:35 +0000

For years pytrends was how Python developers got Google Trends data. Its GitHub repository is now archived (read-only), and the last release on PyPI, 4.9.2, came out in April 2023. Google rate-limits Trends heavily, so pytrends scripts that used to work now stop with HTTP 429 "Too Many Requests" errors, and nobody is fixing the library any more.

This post maps each pytrends call to a maintained alternative, so you can migrate without rewriting your analysis.

Your options

Keep pytrends and slow down. Sleep between requests, retry with backoff and send fewer keywords per payload. That can keep a small script alive, but it breaks again whenever Google changes the internal endpoints pytrends relies on.
The official Google Trends API. Google announced an official API in July 2025. It's an alpha with access by application only, so most projects can't use it yet.
A maintained hosted API. We maintain a Google Trends Scraper on Apify. It paces and retries requests for you and returns the same data pytrends did, as JSON. You call it from Python with Apify's official client.

pytrends calls and their replacements

pytrends	Google Trends Scraper
`build_payload(kw_list, timeframe, geo, cat, gprop)`	Input `searchTerms`, `timeRange`, `geo`, `category`, `searchType`
`interest_over_time()`	`interestOverTime`: `[{date, value, isPartial}]`. Set `"compare": true` to put up to 5 keywords on one 0-100 scale, like a multi-keyword payload
`interest_by_region()`	`interestByRegion`: `[{geoCode, geoName, value}]`
`related_queries()`	`relatedQueries.top` and `relatedQueries.rising`
`trending_searches()`, `realtime_trending_searches()`	`"mode": "trendingNow"` with `trendingGeos`, e.g. `["US", "GB"]`
`related_topics()`, `top_charts()`, `suggestions()`	Not available

Timeframes: now 1-H → past-hour, now 4-H → past-4-hours, now 1-d → past-day, now 7-d → past-7-days, today 1-m → past-30-days, today 3-m → past-90-days, today 12-m → past-12-months, today 5-y → past-5-years, all → since-2004. A custom "2026-01-01 2026-06-30" becomes startDate and endDate.

gprop: "" → web, images, news, froogle → shopping, youtube.

Before and after

A typical pytrends script:

from pytrends.request import TrendReq

pytrends = TrendReq(hl="en-US", tz=360)
pytrends.build_payload(["crm", "erp"], timeframe="today 12-m", geo="US")
df = pytrends.interest_over_time()   # one column per keyword

The same DataFrame from the maintained API (pip install apify-client pandas):

import pandas as pd
from apify_client import ApifyClient  # apify-client 3.x

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("rimulabs/google-trends-insights").call(run_input={
    "searchTerms": ["crm", "erp"],
    "compare": True,                 # one 0-100 scale, like a pytrends payload
    "timeRange": "past-12-months",
    "geo": "US",
    "sections": ["interestOverTime"],
})
rows = [
    {"date": p["date"], "keyword": r["keyword"], "value": p["value"]}
    for r in client.dataset(run.default_dataset_id).iterate_items()
    for p in r["interestOverTime"]
]
df = pd.DataFrame(rows).pivot(index="date", columns="keyword", values="value")

Each keyword record also carries a plain-language summary in insights (average, latest and peak interest, and whether the trend is rising, falling or stable), so you may not need your own smoothing code.

Things that work differently

No session to manage. There's no TrendReq, cookies or proxy settings to look after. Each run is one request from your side.
Batch keywords freely. Without compare, each keyword gets its own report on its own 0-100 scale, which is what most monitoring jobs want.
Values are still relative. Google Trends reports interest from 0 to 100, not search volumes, whichever tool you use.

Cost: $3.00 per 1,000 keyword reports, or $0.50 per 1,000 trending searches. A daily job tracking 20 keywords costs about $1.80 a month.

The full guide, kept up to date, is at rimulabs.com/guides/pytrends-alternative.