MCP App Store
by KeenEthics
Immoprobe icon

Immoprobe MCP Server for Claude and ChatGPT

By Immoprobe

Overview

Immoprobe runs the numbers on a rental property in Germany. Give Claude a listing — a portal link, its pasted text or an exposé PDF — and Claude extracts the purchase price, living space, year built, rent and service charge, looks the address up and reads what open official data says about it: the land value (Bodenrichtwert), social-preservation and redevelopment areas, listed buildings, the rent index location class (Berlin, Hamburg), flood hazard zones, noise maps, the Census 2022 cell and public transport nearby. The calculation covers purchase costs by federal state, financing, depreciation (AfA), income tax, cashflow before and after tax, gross and net yield, the purchase price factor and a ten-year IRR, with every assumption named so you can replace it. Claude can save the listing to your Immoprobe account with its assessment as a note, tell whether a listing is already saved before saving it again, compare candidates side by side and review your portfolio. Nothing is deleted through Claude: removing an object happens on immoprobe.de. Accounts are free, sign-in is by e-mail code, the interface is German and English. Data sources and licences: immoprobe.de/data-sources. Auf Deutsch: Immoprobe rechnet ein Immobilien-Inserat durch, direkt im Chat. Claude liest Kaufpreis, Wohnfläche, Baujahr, Miete und Hausgeld aus dem Inserat, schlägt die Adresse nach und holt offene amtliche Daten dazu: Bodenrichtwert, Milieuschutz und Sanierungsgebiete, Denkmalschutz, Mietspiegel-Wohnlage, Hochwasser- und Lärmkarten, Zensus und ÖPNV. Das Ergebnis: Kaufnebenkosten je Bundesland, Finanzierung, AfA, Steuer, Cashflow vor und nach Steuer, Rendite, Kaufpreisfaktor. Kandidaten lassen sich im Immoprobe-Konto speichern und vergleichen. Kostenloses Konto, Chat auf Deutsch und Englisch.

How to connect

  1. Claude: Open the Immoprobe listing in the Claude connectors directory and choose Connect. Claude listing Immoprobe exposes 15 tools here.
  2. ChatGPT: Open the Immoprobe listing in the ChatGPT app directory and choose Connect. ChatGPT listing Immoprobe exposes 15 tools here.

Tools

add_attachments_from_urls

ChatGPT
Stores files a page links to with a saved object: the listing's photos, the Grundriss (kind floorplan), an exposé PDF. Takes file URLs (not the page), up to 20 per call; Immoprobe fetches them itself and never stores the same URL twice. Without kind an image is a photo and anything else a document; the first photo is the cover. What fails (a login wall, a bot check, a type or size outside the limits) comes back in failed with the reason, and a retry answers the same. A file on the user's machine cannot come through a tool: the object's page on the website takes it. Example arguments: {"objectId":"obj_…","items":[{"url":"https://pictures.immobilienscout24.de/listings/abc-1392x1044.jpg","kind":"photo"},{"url":"https://pictures.immobilienscout24.de/listings/grundriss-1392x1044.png","kind":"floorplan","name":"Grundriss.png"}]}

add_note

ChatGPT
Attaches a note to a saved object: kind analysis for an assessment of the listing (what to check, risks, questions), viewing for a Besichtigung, question for the Makler. Markdown, up to 20 000 characters. Notes are append-only: a correction is a further note. idempotencyKey (any string unique to the note) makes a retried call safe: the same key adds nothing twice.

calculate_property

ChatGPT
Calculates a German residential investment property: cashflow before and after tax, yields, the purchase price factor, AfA, a ten-year IRR, a 30-year projection and a verdict. Required: purchasePrice (EUR), livingArea (m²), stateCode, coldRent (EUR per month, cold). Every other field has a default; the defaults that were used come back in assumed. Financing is equity (EUR) or ltv (loan in % of the price, 80 by default); kfwAmount is a KfW tranche within that loan. A new build bought before completion takes handover (YYYY-MM) and, once signed, contractMonth. landValuePerSqm (the Bodenrichtwert) with plotArea, coOwnershipShare and coOwnershipTotal sets the land share by the BMF Arbeitshilfe; without them the land share is a default. The answer is the calculation as text in the requested language, the structured result, and shareUrl, the same calculation with charts in the browser. Example arguments: {"purchasePrice":289000,"livingArea":62,"yearBuilt":1968,"stateCode":"NW","coldRent":1290,"serviceCharge":240,"hasBroker":true}

check_known_listings

ChatGPT
Says which listings of a results page are already saved: takes their ids (up to 200) with the source portal and answers, per id, known (with object id, status and the last saved price) or new.

clone_object

ChatGPT
Saves a copy of a saved object under a new title with some figures changed, so that variants can be compared: another financing (equity, ltv, interestRate, kfwAmount …) or a negotiated purchasePrice. input holds only the fields that change; null returns a field to its default and lists it in assumed. Address, the other figures, source, listing page and lists are copied; notes, links, files and the portal's externalId stay with the original. Counts as a new object; the answer carries the copy's url. Example arguments: {"objectId":"obj_1a2b3c","title":"Malplaquetstraße 27 · 30 % EK, 15 Jahre fest","input":{"equity":120000,"fixedRateYears":15}} {"objectId":"obj_1a2b3c","title":"Malplaquetstraße 27 · 90 % Beleihung","input":{"equity":null,"ltv":90,"interestRate":3.9},"assumed":["interestRate"]}

compare_objects

ChatGPT
Puts two to five saved objects side by side: one row per metric (price, price per m², yields, cashflow before and after tax, return on equity, LTV, depreciation …), values in the order of the objects, and bestIndex marking the winner where high or low is clearly better. The answer carries the url of the same table on the website.

find_objects

ChatGPT
Lists the user's saved properties, newest first, as compact rows with KPIs, status, lists and note count. Filters: status (active, discarded), a list name, the source portal, or free text over title and address. Objects in the trash (discarded) appear only when that status is requested. When truncated is true, nextCursor continues the list as cursor.

get_account

ChatGPT
The Immoprobe account this connection acts for: email, plan, the remaining quotas for saved objects, address lookups in the current month and attachment storage, and what this credential may do.

get_guide

ChatGPT
The guide to working with Immoprobe: the field mapping from a listing to the calculator with units and defaults, what counts as an assumption, how each data layer of the address lookup is read, how notes and saved objects are structured, and the conduct expected on listing portals. Answers the guide as text in the requested language.

get_object

ChatGPT
One saved property in full: the calculator input as saved, which fields are assumptions, headline KPIs, every note, the links to its other pages, the files attached to it (names and kinds; the bytes stay in the web app) and who saved it.

lookup_address

ChatGPT
Finds where an address is and what the data layers say at that point: the Bodenrichtwert with its date and GFZ, statutory designations (Milieuschutz, Sanierungsgebiet, Denkmal), the Zensus cell, flood hazard, the Mietspiegel Wohnlage with the Vergleichsmiete cell (when yearBuilt and livingArea are given), noise, public transport and the cadastral parcel. Takes lat and lng, or the address as the listing writes it (a query or structured fields). An ambiguous address answers candidates, each with a candidateId that a second call resolves. approximate: true means the house number was unknown and the values stand for the street. A layer missing from the answer has no data at the point. attribution names the sources of the values with their licences. One address lookup counts against the monthly quota. Example arguments: {"lat":52.548,"lng":13.36} {"query":"Malplaquetstraße 27, 13347 Berlin"} {"candidateId":"51a2…"}

save_object

ChatGPT
Saves a listing as an object in the user's account: the calculator input (the same fields as the calculation), the address as the listing shows it (formatted is enough; postalCode, street, houseNumber, lat, lng and precision when known), source and the portal's externalId plus listingUrl, and in the same call a first note with the assessment and the lists to file it in. assumed names the fields that were guessed rather than read from the listing. The same source and externalId updates the existing object instead of creating a second one. The object is in the user's list at once, marked as saved by an assistant. The answer carries the url and possibleDuplicates, the saved objects at the same address (nothing is merged). Example arguments: {"address":{"formatted":"Malplaquetstraße 27, 13347 Berlin","lat":52.548,"lng":13.36,"precision":"house"},"input":{"purchasePrice":395000,"livingArea":74,"yearBuilt":1910,"stateCode":"BE","coldRent":1250,"serviceCharge":310,"hasBroker":true},"assumed":["serviceCharge"],"source":"immoscout24","externalId":"123456789","listingUrl":"https://www.immobilienscout24.de/expose/123456789","lists":["Berlin"],"note":{"kind":"analysis","body":"Altbau 1910, Erbbaurecht nicht erwähnt — nachfragen.\n\nFragen an den Makler:\n- Höhe der Instandhaltungsrücklage?","idempotencyKey":"expose-123456789-analysis"}}

set_object_lists

ChatGPT
Puts a saved object into lists by name — Berlin, Vielleicht, Besichtigen — creating names that do not exist yet. The names replace the object's whole membership. Answers the lists with their object counts.

set_object_status

ChatGPT
Moves a saved object into the trash (discarded: out of the lists, no longer counted against the plan, still reported as known by the listings check so the same listing is not saved again, deleted for good after 90 days) or back to the list (active). Answers the object.

add_attachments_from_urls

Claude

add_note

Claude

calculate_property

Claude

check_known_listings

Claude

clone_object

Claude

compare_objects

Claude

find_objects

Claude

get_account

Claude

get_guide

Claude

get_object

Claude

lookup_address

Claude

save_object

Claude

set_object_lists

Claude

set_object_status

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/immoprobe"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/immoprobe)