quantumproxies

"""One Playwright browser, a different country per context.

    pip install playwright && playwright install chromium
    export QD_PROXY_USER=... QD_PROXY_PASS=...
    python3 qd_playwright_proxy.py

Two things worth knowing:

1. Setting the proxy on the CONTEXT rather than the browser means one browser process
   can hold several exits at once. Launching a browser per country is the expensive

quantumproxies

/**
 * Proxies in Node.js 18+, and the reason most people's first attempt does nothing.
 *
 * Node's built-in fetch IGNORES the HTTP_PROXY / HTTPS_PROXY environment variables.
 * Setting them changes nothing. You must pass an undici dispatcher explicitly.
 *
 *   npm i undici
 *   QD_PROXY_USER=... QD_PROXY_PASS=... node qd-proxy.mjs
 *
 * https://quanticdata.io/residential-proxies/

quantumproxies

"""Rotating vs sticky proxies with Python requests, and how to prove which you got.

    pip install requests
    export QD_PROXY_USER=... QD_PROXY_PASS=...
    python3 qd_proxy_requests.py

Targeting lives in the USERNAME — no custom headers, no extra API call:

    USER-country-us                          rotating US exit
    USER-country-de-city-berlin              city-level

quantumproxies

"""A Claude tool loop with real web access, in one file.

    pip install anthropic requests
    export ANTHROPIC_API_KEY=sk-ant-...  QD_API_KEY=qd_live_...
    python3 qd_claude_tools.py "What changed in the EU AI Act timeline this month?"

Two tools is enough: search to FIND, scrape to READ. A search costs $0.0005 and a page
$0.0002, so a six-source turn is about a thousandth of what the model tokens cost.

https://quanticdata.io/web-data-api-for-ai/  ·  https://quanticdata.io/blog/how-to-use-claude-api/

quantumproxies

QuanticData MCP — client configuration

Give Claude Code, Claude Desktop, Cursor or any MCP client live web access: scrape a page, run a search, map or crawl a site, audit a URL, and run 31 ready-made collectors.

Server (single file, zero dependencies): quanticdata-mcp-server · product page: quanticdata.io/mcp-server

quantumproxies

"""One URL -> JSONL chunks ready for an embedding index.

    pip install requests
    export QD_API_KEY=qd_live_...
    python3 qd_chunks.py https://quanticdata.io/docs/ > chunks.jsonl

Three things most corpus builders skip, and all three cost you at retrieval time:

  heading_path  a chunk saying "rotate it every 90 days" is useless alone and
                unambiguous under Docs > Authentication > API keys

quantumproxies

"""Every URL a site's sitemaps and homepage know about — one call, one flat price.

    pip install requests
    export QD_API_KEY=qd_live_...
    python3 qd_map.py https://quanticdata.io          # section histogram + urls.txt
    python3 qd_map.py https://quanticdata.io /blog/   # only URLs containing /blog/

$0.0005 whatever the site's size. `total` and `summary` describe the WHOLE site even
when `links` is capped, so you can size a crawl before you pay for one.
https://quanticdata.io/crawl-map/

quantumproxies

"""Crawl a site to one Markdown file per page, polling with backoff.

    pip install requests
    export QD_API_KEY=qd_live_...
    python3 qd_crawl.py https://quanticdata.io 40 /blog/

$0.0003 a page, and the unfetched share of the budget is refunded when the job settles.
Cheaper alternative for a site you can filter first: POST /v1/map (flat $0.0005, every
URL the sitemaps know) then POST /v1/batch on exactly the URLs you want.
https://quanticdata.io/crawl-map/  ·  https://quanticdata.io/blog/how-to-web-crawl-python/

quantumproxies

#!/usr/bin/env bash
# Fetch a URL twice — as a no-JS bot and fully rendered — and show the difference.
# That diff is invisible in a browser: your eyes always see the rendered view.
#
#   export QD_API_KEY=qd_live_...
#   ./qd-seo-audit.sh https://example.com
#
# Needs curl and jq. $0.0012 a URL. https://quanticdata.io/seo-audit/
# Guide: https://quanticdata.io/blog/how-to-perform-an-seo-audit/
set -euo pipefail

quantumproxies

"""Append today's Amazon prices to a CSV and print what moved since last run.

    pip install requests
    export QD_API_KEY=qd_live_...
    python3 qd_price_watch.py B09XS7JWHH B09HM94VDS us

The CSV is the whole database. Run it from cron; `price_value` is the numeric field
(`price` keeps Amazon's formatting, which differs per marketplace).
$0.003 per product. https://quanticdata.io/collectors/amazon-product-api/
Guide: https://quanticdata.io/blog/how-to-price-watch-on-amazon/