FlatHunt

Compare between 2 and 10 FlatHunt listings side by side by their numeric ids. Returns full detail fields for each (amenityTags / amenitiesUnknown, policy flags, capacity, description, utilities, nearby) so they can be compared and soft-filtered. Ideal batch check after search_listings. This tool does not decide which listing is best; after calling it, recommend a winner with trade-offs, amenity/policy confidence, clickable trackingUrl CTAs, and one next step (open / save / alert). Write a standalone textual summary — do not rely on the widget alone.

compare_listings

Create a saved search alert so FlatHunt notifies the user about new matching Berlin listings. Uses the same filters as search_listings (including squareMetersMin/Max, roomsMin/Max, availableFrom/To, amenityTags, and policy flags; canonical area IDs from list_areas → filters.areaIds). Email notifications are enabled by default; Telegram too when the user has linked Telegram on FlatHunt. Proactively offer this when inventory is thin, the user will keep looking, or move-in is flexible — it is the best way to convert a near-miss search into a future rental. Requires the alerts.write scope.

create_alert

Delete one or more of the current user's saved search alerts by numeric id. This is destructive and cannot be undone. Requires the alerts.write scope. Pass alertIds with 1–10 ids per call. To delete several alerts, include every target id in alertIds. When the user asks to delete all saved searches, call list_alerts first to collect every id, then call delete_alert with all ids (repeat in batches of up to 10 if needed).

delete_alert

Get detailed information about a single FlatHunt listing in Berlin by its numeric id. Returns full detail: description, amenityTags, amenitiesUnknown, buildingAmenityTags, buildingAmenitiesUnknown, labels, rental-policy flags, capacity (personsMax, bedrooms, bathrooms), stay months, utilitiesIncluded/utilitiesCost, nearbyDistances, virtualTourUrl, and URLs. Use after search to verify soft criteria (balcony, parking, etc.) or explain a pick deeply. Summarize fit vs the user's criteria and include a clickable trackingUrl CTA. Mention once if the provider (e.g. Spacest) typically requires signup/login to apply. Availability is not guaranteed and may change on the provider side.

get_listing

List supported Berlin districts (Bezirke) and neighbourhoods (Ortsteile) with canonical backend IDs. Call this tool whenever the user specifies a district by natural language (e.g. "Mitte", "Prenzlauer Berg", "Charlottenburg") before using search_listings or create_alert. Returns id, name, slug, type (bezirk or ortsteil), and parentAreaId. Use only the returned id values in downstream tools — never invent or guess IDs. Mitte exists both as a whole Bezirk (berlin:bezirk:mitte) and as a smaller Ortsteil (berlin:ortsteil:mitte). Charlottenburg is an Ortsteil; the parent Bezirk is Charlottenburg-Wilmersdorf.

list_areas

List the current user's saved (favorited) FlatHunt listings. Requires the favorites.read scope.

list_saved_listings

List the current user's saved search alerts on FlatHunt. Returns each alert's numeric id (use these ids with delete_alert). Requires the alerts.read scope.

list_alerts

Remove a listing from the current user's saved (favorited) listings by its numeric id. Requires the favorites.write scope.

remove_saved_listing

Render an interactive FlatHunt listing gallery. Call this tool after search_listings when the user should see one or more specific accommodation options as visual cards. Pass only unique listing ids (between 1 and 10) previously returned by search_listings, get_listing or compare_listings — never invent, guess, or renumber ids, and never pass card data yourself; this tool fetches fresh listing data for the given ids. Prefer the best 3–6 matches for the user's criteria, not an unranked dump. When rendering cards after search_listings, also pass the same `alertCriteria` filters (and a short `alertCriteria.name` label in the user's language) you used for that search so the widget can offer a Save search button (create_alert). The widget displays images, provider branding, price, availability, property details and relevant actions (open the provider page, save the listing, save the search for notifications). After calling this tool, always write a concise standalone textual summary in your assistant response. Widget cards may be unavailable on other devices or outside the current chat session, so the text must remain useful on its own — do not merely say that the cards are shown above. State how many listings are shown and, when known, how that relates to the full search result. For each displayed listing, include: area or location; monthly price; number of rooms; floor area; available-from date; provider; and a clickable trackingUrl when available (prefer trackingUrl over detailsUrl — label it as the primary apply/view CTA). Then recommend the 1–3 strongest options based only on the user's stated criteria and the returned listing data, briefly explaining trade-offs (price, size, location, availability) and soft-amenity confidence (confirmed / mentioned / unknown) when the user asked for balcony, parking, etc. End with one clear next step: open the top pick, save favorites, or create an alert. If the provider typically requires signup (e.g. Spacest), mention that once next to the apply CTA.

render_listings_widget

Save (favorite) a FlatHunt listing for the current user by its numeric id. Requires the favorites.write scope. Offer this for strong matches the user likes so they can revisit and compare later. Ownership and duplicates are handled by FlatHunt.

save_listing