DEV Community: Krzysztof Filipiak

I built an API for Polish tax rulings: 550k+ official interpretations for AI agents

Krzysztof Filipiak — Thu, 22 Oct 2026 06:40:17 +0000

I built an API for Polish tax rulings: 550,000+ official interpretations from the Ministry of Finance's EUREKA database as structured JSON – thesis, outcome, key sections and full Markdown. It works as an MCP tool for Claude and ChatGPT via Apify. Pay per result ($0.001).

👉 apify.com/kris_fili/poland-tax-interpretations

Why

In Poland, individual tax interpretations issued by the National Tax Information (KIS) are a core source for tax advisors and accountants – more than 525,000 of them are public in EUREKA. But EUREKA only offers a website search, which is hard to use from code or from an AI agent.

What you get

Search by phrase (in Polish), signature or document ID
Document types: individual interpretations, binding VAT rate information (WIS), general interpretations, tax explanations, court rulings and more
For each document: signature, issue date, status, thesis and a link to the official source
With details: the outcome (was the taxpayer's position correct?), the key sections (facts, question, taxpayer's position, assessment) and the full text as Markdown
For WIS: the VAT rate, classification, legal basis and validity period as separate fields

Example

Input:

{
  "query": "leasing samochodu osobowego",
  "detailLevel": "sections",
  "maxResults": 10
}

Output (shortened):

{
  "signature": "0114-KDIP4-2.4012.626.2026.3.MC",
  "issueDate": "2026-09-25",
  "thesis": "Prawo do odliczenia podatku VAT od wydatków związanych z samochodem osobowym.",
  "outcome": "incorrect",
  "sections": { "question": "…", "applicantPosition": "…", "assessment": "…" },
  "url": "https://eureka.mf.gov.pl/informacje/podglad/712074"
}

Use it from an AI agent (MCP)

Add the Actor through the Apify MCP server:

https://mcp.apify.com/?actors=kris_fili/poland-tax-interpretations

Then ask, for example: "Find recent Polish tax interpretations on VAT deduction for leased cars and quote the signatures."

Data source

Data comes live from the public EUREKA service of the Polish Ministry of Finance. Official documents are not protected by copyright in Poland, and the Ministry anonymises them before publication. The project is independent and not affiliated with the Ministry. Results are information, not tax advice.

How much apartment does 600,000 PLN buy in Poland's largest cities?

Krzysztof Filipiak — Mon, 19 Oct 2026 06:40:17 +0000

Asking prices on listing portals tell you what sellers hope for. Notarial deeds tell you what buyers actually paid. Here is how much apartment a fixed budget buys in each of Poland's largest cities, using the median price per m² from deeds signed in the last 12 months.

The numbers

City	Median price per m² (PLN)	m² for 600,000 PLN
Czestochowa	7,243	83
Lodz	9,250	65
Olsztyn	9,343	64
Kielce	9,483	63
Torun	9,512	63
Rzeszow	9,613	62
Bialystok	9,937	60
Katowice	10,000	60
Lublin	10,129	59
Szczecin	10,561	57
Poznan	11,540	52
Gdynia	11,936	50
Gdansk	12,622	48
Wroclaw	12,862	47
Krakow	14,961	40
Warsaw	15,742	38

What stands out

Czestochowa gives you the most space: 83 m².
Warsaw the least: 38 m² – 2.2× less than Czestochowa.

Methodology

Apartment sales from the Polish Real Estate Price Register (RCN, published by GUGiK; free since 13 February 2026): market sales of full ownership only, implausible prices removed.
12-month window ending at the newest deed published in each city – counties publish with different delays.
Weekly open data for all cities (CSV, CC BY 4.0) is on GitHub: poland-housing-prices.

Get the data for any address

Transaction prices from notarial deeds – median, quartiles, trend and the deeds around any address.
Apartment valuation (AVM) – an estimate for any apartment from comparable deeds.
New-build prices – developers' asking prices from their daily price lists.

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("kris_fili/poland-real-estate-transaction-prices").call(run_input={
    "location": "Kraków, Rynek Główny 1",
    "radiusMeters": 1000,
    "monthsBack": 24,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    if item["recordType"] == "summary":
        print(item["statistics"]["pricePerM2Pln"])

The same tools work inside AI assistants through MCP. Questions or ideas for another analysis? Leave a comment.

How to get real apartment transaction prices in Poland via API (official RCN data)

Krzysztof Filipiak — Tue, 06 Oct 2026 06:40:31 +0000

Since 13 February 2026 the Polish Real Estate Price Register (Rejestr Cen Nieruchomości, RCN) is free and public. Every notarial deed of a property sale ends up there: the price, the floor area, the number of rooms, the floor, primary or secondary market and the location. For anyone analysing the Polish housing market this is the first official source of prices actually paid, not asking prices from listing portals.

For example, for the whole of Krakow the register currently gives 11,063 apartment sales in the 12 months up to September 2026, with a median of 14,961 PLN/m² (new builds 15,000, resale 14,872).

Why the raw register is hard to use

The data is published through an OGC WFS service that returns GML in the Polish coordinate system EPSG:2180. To get something useful you have to:

build XML filters (by county TERYT code or by a bounding box),
page through results 1,000 records at a time,
convert coordinates to GPS,
remove non-market sales (municipal flats sold with discounts), partial shares and parking spaces sold as shares of a garage hall,
handle units that differ between counties and outliers caused by typos in deeds,
remember that each county publishes with a different delay (from weeks to months).

The quick way: one API call

I wrapped all of this into an Apify Actor: Poland Real Estate Transaction Prices. You give it an address, a city or coordinates and get a statistical report (median price per m², quartiles, quarterly trend, primary vs secondary market, year-on-year change) plus the list of transactions.

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("kris_fili/poland-real-estate-transaction-prices").call(run_input={
    "location": "Kraków, Rynek Główny 1",
    "radiusMeters": 1000,
    "monthsBack": 24,
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    if item["recordType"] == "summary":
        print(item["statistics"]["pricePerM2Pln"])

It also works without code: open the Actor page, type an address and press Start. Results can be downloaded as Excel, CSV or JSON, or used from Make, Zapier, n8n or an AI agent through MCP.

Asking prices and the gap

Since 11 September 2025 Polish developers must publish their price lists every day on dane.gov.pl. A second Actor, Poland New-Build Apartment Prices, collects all of them every morning (hundreds of thousands of offers), cleans dozens of different file formats and geocodes the addresses.

The third one, Poland Housing Price Gap, compares both sources for any location: how much more developers ask today than buyers paid in recent notarial deeds.

Good to know

Prices from deeds of new apartments often lag behind current price lists, because deeds are usually signed at handover.
A few counties do not publish data yet – every report shows the date of the latest deed found.
No personal data is processed: the register does not contain names of buyers or sellers.

Questions or ideas for other Polish open-data APIs? Let me know in the comments.