Feedsmith

Reliable data feeds from public sources, kept running.

When Apify, Octoparse or a ChatGPT script gets blocked and returns empty pages, I take over the extraction and keep it alive.

Get a free feasibility check → See a real output

Two minutes. Send the public URL and the fields you need. I answer with what is extractable, how, and at what price.

raw public web → structured feed live
title price availability rating source fetched_at 0000 rows delivered

Offers

Start with an audit. Scale to a maintained feed.

Four ways in, from a low-risk check to a feed I keep alive.

Hidden-API Audit

See details →

I check one public site, find the cleanest data route, and send you about 50 real sample rows. The low-risk way to start.

€150fixed fee

One-off Scraper

See details →

A robust scraper for one public source, delivered as tested code you own. Clean CSV, JSON or API output.

€600to 1,500, one time

Managed Data Feed

See details →

Scraper, scheduled delivery and monitoring. The source changes, it breaks, I fix it, usually before you notice.

€500setup, then €150 to 500 / month

Discord Bot

See details →

A custom bot for your community: automation, public-data lookups, scheduled posts. Built and hosted.

€500to 2,500, plus hosting

From a blocked scraper to a feed you can rely on.

1

Audit

I check the public source and find the most reliable extraction route, a hidden API where one exists.

2

Build

A tested scraper that handles JS-heavy and Cloudflare-protected public pages, outputting clean data.

3

Schedule & monitor

Delivery on your cadence, with monitoring. When the source changes and it breaks, I repair it.

4

You operate & own

You run the tool and own the data. I build and maintain it: a tool, not an operating service.

The deliverable

A dataset you can check, not a promise.

Real output from the open-source starter behind the Managed Data Feed offer, run against a public sandbox catalogue. Same shape as what lands in your inbox or your database.

books-demo.csv · run 2026-06-17 21:05 UTCCSV
title price availability rating source
A Light in the Attic£51.77In stockThreebooks.toscrape.com
Tipping the Velvet£53.74In stockOnebooks.toscrape.com
Soumission£50.10In stockOnebooks.toscrape.com
Sharp Objects£47.82In stockFourbooks.toscrape.com
Sapiens: A Brief History of Humankind£54.23In stockFivebooks.toscrape.com
Public sandbox source. Factual, non-personal fields. One row per product, one column per field, with the fetch timestamp kept.

Your feed carries your fields and nothing else: the columns you asked for, in the format you asked for.

Scope

Reliability, not circumvention.

  • Public, logged-out pages only. Factual and business data: prices, specs, stock, listings, market data, schedules. No logins, no paywalls, no accounts.
  • No personal data. Strictly non-PII. No names with contacts, profiles or photos. B2B and factual data only.
  • robots.txt and rate limits respected. Pages are read like a normal visitor would, at a respectful rate. Output is structured and transformed.
  • You operate and own the data. I build and maintain the tool, you run it and control the data. A clean, documented handover every time.

Typically in scope

Product prices, specs, stocke-commerce
Listings (price, size, location)real estate
Catalogue and marketplace datamarketplace
Market data, schedules, pricestravel · finance
Feedsmith builds a tool for reading public, factual data. It is not used to access logins, paywalls or personal data. You remain the operator and owner of the data, and confirm your right to access the source.

Proof

Open source, tested, and runnable.

Not a slide deck. Real code you can read, clone and run: the hard 20% of scraping that no-code tools fail at.

feeds/books_demo.yamlconfig
id: books-demo
source: books.toscrape.com
fields: [title, price, availability, rating]
rate_limit_seconds: 1.5
urls: ["https://books.toscrape.com/catalogue/page-1.html"]
schedule:
  interval_seconds: 30
output:
  kind: csv
  path: data/books-demo.csv
One feed, one file. That is the whole configuration behind the sample above.
starter

managed-data-feed-starter

The Managed Data Feed offer in code: resilient fetch, no-PII policy, schedule, self-healing monitor, CSV / JSON / webhook output.

View on GitHub →
template

hidden-api-extraction-template

Read a site's internal JSON API directly instead of parsing fragile HTML: faster, and far more stable.

View on GitHub →
demo

tls-fingerprint-scraper-demo

Why a public page returns empty: a TLS fingerprint check before render. How to read it like a normal browser, compliantly.

View on GitHub →

The why behind the work.

Why your no-code scraper keeps breaking

Layout changes, JS-rendered data, fingerprint checks, and what actually fixes them.

Read the guide →

What is a hidden API

The internal JSON endpoints behind a page, and why calling them beats HTML parsing.

Read the guide →

Cloudflare-protected public sites

Empty pages are usually a TLS fingerprint check rather than a CAPTCHA, and there is a compliant way to read them.

Read the guide →

Get a free feasibility check

Tell me the public site and the data you need. I will tell you if it is doable, how, and in which language.

Or email hello@feedsmith.net · Based in Switzerland · DE / FR / IT / EN.