Skip to main content
Glama

kilid-mcp: Kilid (کیلید) real-estate MCP server for AI agents, homes for sale and rent in Iran

kilid-mcp is an unofficial, read-only MCP server that lets an AI agent search homes for sale and rent, read house prices per square metre and look up agencies on Kilid (کیلید, kilid.com), a large Iranian real-estate portal. Its 8 tools search buy and rent listings with price, area, room and district filters, open one listing in full, read the median price per m² by city, district and neighborhood with a 12-month trend, rank neighborhoods by price, and find agencies. No API key or account is needed; it works in Claude, Cursor, Copilot and any other MCP client.

PyPI Python CI License: MIT

For AI agents · Quick start · Tools · Limits · فارسی


For AI agents

  • What: unofficial MCP server (model context protocol) for kilid.com: cities and districts, listing search (sale and rental), listing detail, house-price statistics with a 12-month trend, neighborhood price ranking, agencies.

  • Install: uvx kilid-mcp or claude mcp add kilid -- uvx kilid-mcp. Transport: stdio. Needs uv (Python 3.10+).

  • Auth: none. Optional KILID_MCP_PROXY (HTTP proxy URL, or a comma-separated list); system proxy settings are ignored.

  • Access: read-only. It cannot log in, save listings, contact agencies or reveal phone numbers, so no confirmation is needed before calling any tool.

  • Tools: 8, all prefixed kl_. Typical sequence: kl_cities -> kl_districts -> kl_search_listings -> kl_listing; kl_price_stats and kl_neighborhood_prices for market levels; kl_agencies -> kl_agency.

  • Use when: the user wants apartments or houses for sale or rent in an Iranian city, what a square metre costs in a district, or the priciest neighborhoods.

  • Do not use when: the user needs advertiser phone numbers, sold-price records, or to publish an ad.

  • Output: compact JSON. Prices are Toman. Rentals return the deposit as price_toman and the monthly rent as rent_toman. Text is Persian.

Related MCP server: otaghak-mcp

Quick start

You need uv. No API key or account.

Claude Code:

claude mcp add kilid -- uvx kilid-mcp

Claude Desktop, Cursor, VS Code and other clients:

{
  "mcpServers": {
    "kilid": { "command": "uvx", "args": ["kilid-mcp"] }
  }
}

Example prompts:

  • "Find 2-bedroom flats to buy in Tehran between 60 and 100 m²."

  • "What is the median price per square metre in Tehran and in its first district?"

  • "Which Karaj neighborhoods are the most expensive?"

  • آپارتمان اجاره‌ای دو خوابه در شیراز پیدا کن.

  • قیمت هر متر خانه در ولنجک چقدر است؟

Real answer (17 Nov 2026): kl_price_stats(area="tehran") returned a median of 237,179,487 Toman per m² (period 1405-08, sample 433,751 listings), and منطقه 1 at 571,428,571 Toman per m².

Tools

Tool

What it does

kl_cities

Find a city id and slug by Persian name, with whether district data exists.

kl_districts

List municipal districts or neighborhoods of a city with their ids and slugs.

kl_search_listings

Search buy or rent listings by city, districts, property type, price, rent, area, rooms, sort and page.

kl_listing

Read one listing: price, area, rooms, floor, facilities, description, map point and agency.

kl_price_stats

Median price per m² for a city or area, its sub-areas and the 12-month trend.

kl_neighborhood_prices

Rank a city's neighborhoods by median price per m², with a data-quality flag.

kl_agencies

Search agencies registered on Kilid by name.

kl_agency

Read one agency: address, district, location, active listings and agent count.

Limits

  • Prices are asking prices from listings, not sale records. year_change_pct is the site's own yoyPct field; the page itself shows a client-side figure that differs a little.

  • Featured (paid) listings can appear on top even when they fall outside the filters. rent_toman 0 means deposit only.

  • Advertiser phone numbers sit behind a reveal button on the site and are deliberately not returned.

  • Kilid is a Next.js site and several tools read data embedded in its server-rendered pages. If Kilid renames those props, a tool can come back empty until the server is updated.

  • Heavy use and rate limits were not probed. Requests run two at a time with a short cache. It answers from abroad and from Iran.

How it works

Kilid has no public API. The server calls the JSON routes the site's own pages use (/api/search, /api/listings, /api/regions, /api/provinces, /api/neighborhoods/finder, /api/agency/suggest) and reads the data embedded in its server-rendered pages (listing page one, listing detail, price reports, agencies). The routes were mapped in a real browser and every tool's output was checked in Chrome against the rendered kilid.com pages. Requests present Chrome's TLS fingerprint.

فارسی

kilid-mcp یک سرور MCP غیررسمی و فقط‌خواندنی است که به دستیار هوش مصنوعی اجازه می‌دهد آگهی‌های خرید و اجاره‌ی ملک، قیمت هر متر مربع خانه و بنگاه‌های املاک را در کیلید (kilid.com) جست‌وجو کند. هشت ابزار دارد: جست‌وجوی آگهی با فیلتر قیمت، متراژ، تعداد اتاق و محله، دیدن کامل یک آگهی، قیمت میانگین هر متر به‌تفکیک شهر و منطقه و محله، رتبه‌بندی محله‌ها و پیدا کردن بنگاه. کلید API یا حساب کاربری لازم نیست. نصب در Claude Code:

claude mcp add kilid -- uvx kilid-mcp

Development

uv sync
uv run ruff check . && uv run ruff format --check .
uv run python -m pytest -q          # offline, saved fixtures
uv run python -m pytest -m live -q  # real kilid.com

Releases: bump the version in pyproject.toml and server.json, run uv lock, then push a v* tag; GitHub Actions publishes to PyPI and the MCP Registry.

Disclaimer

Unofficial and not affiliated with or endorsed by Kilid. It is read-only and uses the public pages and routes of kilid.com, which can change without notice.

Author

Built and maintained by Sepehr Radmard (سپهر رادمرد). LinkedIn · GitHub · more servers in iran-mcp.

License

MIT

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      B
      maintenance
      Gives AI agents read-only access to Bulgarian real-estate data from imotai.bg, letting them list districts, search and fetch active listings, pull district-level market statistics, and estimate values from current asking prices. All results are field-allowlisted and rate limited, with no credentials, broker contact details, or written descriptions exposed.
      5
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI agents to search Iranian villas, cottages, and apartments, check nightly availability, calculate exact stay totals for dates and guests, and read reviews and host statistics. It is a read-only stdio MCP server with no login, booking, payment, or SMS endpoints.
      15
      145 PyPI
      2
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI agents to search and read Divar classified ads across Iranian cities, covering homes, cars, phones and more, with price statistics, deal finding, fair-price appraisal, ad comparison and map browsing. It is read-only, needs no account or API key, and returns compact JSON with prices in Toman.
      8
      31
      296 PyPI
      1
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Enables AI agents to search Iranian places, read place details, reviews, photos and menus, find nearby places, reverse geocode coordinates, check air quality, and get car, motorcycle, bicycle, walking or bus/metro directions, travel times and Tehran traffic-zone status without an API key.
      11
      144 PyPI
      MIT