UK Property Data
UK property data MCP server — Land Registry comps, EPC, Rightmove, rental yields, stamp duty, Companies House. 13 tools.
README
Property Shared
<!-- mcp-name: io.github.paulieb89/property-shared -->
UK property data in one package. Pulls Land Registry sales, EPC certificates, Rightmove listings, rental yields, stamp duty calculations, planning portal links, and Companies House records.
Use it as a Python library, CLI, HTTP API, or MCP server for AI agents.
What You Get
| Data Source | What It Returns |
|---|---|
| Land Registry PPD | Sold prices, dates, property types, area comps with median/percentiles |
| EPC Register | Energy ratings, floor area, construction age, heating costs |
| Rightmove | Current listings (sale + rent), prices, agents, listing details |
| Yield Analysis | Gross yield from PPD sales + Rightmove rentals combined |
| Stamp Duty | SDLT calculation with April 2025 bands, BTL surcharge, FTB relief |
| Block Analyzer | Groups flat sales by building to spot investor exits |
| Planning | Local council planning portal lookup (99 verified councils) |
| Companies House | Company search and lookup by name or number |
Skills
Want structured property reports instead of raw data? Claude skills that chain these tools into investment summaries are available at bouch.dev/products.
Install
pip install property-shared
# or with uv
uv add property-shared
Extras: [mcp] for MCP server, [cli] for CLI, [api] for HTTP server, [dev] for tests.
pip install property-shared[mcp,cli]
# or
uv add property-shared --extra mcp --extra cli
Use as a Python Library
from property_core import PPDService, calculate_yield, calculate_stamp_duty
# Get comparable sales for a postcode
comps = PPDService().comps("SW1A 1AA", months=24, property_type="F")
print(f"Median flat price: {comps.median:,}")
# Calculate rental yield
import asyncio
result = asyncio.run(calculate_yield("NG1 1AA", property_type="F"))
print(f"Gross yield: {result.gross_yield_pct}%")
# Stamp duty
sdlt = calculate_stamp_duty(250000, additional_property=True)
print(f"SDLT: {sdlt.total_sdlt:,.0f} ({sdlt.effective_rate}%)")
All models are available at top level:
from property_core import (
PPDTransaction, PPDCompsResponse, EPCData,
RightmoveListing, RightmoveListingDetail,
PropertyReport, YieldAnalysis, RentalAnalysis,
BlockAnalysisResponse, CompanyRecord, StampDutyResult,
)
Interpretation helpers (core returns numbers, you decide how to label them):
from property_core import classify_yield, classify_data_quality, generate_insights
Use as CLI
pip install property-shared[cli] # or: uv add property-shared --extra cli
# Comparable sales
property-cli ppd comps "SW1A 1AA" --months 24 --property-type F
# Rental yield
property-cli analysis yield "NG1 1AA" --property-type F
# Stamp duty
property-cli calc stamp-duty 300000
# Rightmove search (with sort)
property-cli rightmove search-url "NG1 1AA" --sort-by most_reduced
# Full property report
property-cli report generate "10 Downing Street, SW1A 2AA" --property-type F
Add --api-url http://localhost:8000 to any command to route through the HTTP API instead of calling core directly.
Use as MCP Server (AI Agents)
For Claude.ai, Claude Code, ChatGPT, or any MCP-compatible host.
pip install property-shared[mcp] # or: uv add property-shared --extra mcp
property-mcp # starts stdio transport
12 tools available: property_report, property_comps, ppd_transactions, property_yield, rental_analysis, property_epc, rightmove_search, rightmove_listing, property_blocks, stamp_duty, planning_search, company_search.
Remote server deployed at https://property-shared.fly.dev/mcp (Streamable HTTP).
See mcp_server/README.md for connection setup and tool details.
Use as HTTP API
pip install property-shared[api] # or: uv add property-shared --extra api
property-api # starts on port 8000
Interactive docs at http://localhost:8000/docs.
Key endpoints:
GET /v1/ppd/comps?postcode=SW1A+1AA&property_type=F&enrich_epc=trueGET /v1/analysis/yield?postcode=NG1+1AA&property_type=FGET /v1/analysis/rental?postcode=NG1+1AA&purchase_price=200000GET /v1/rightmove/search-url?postcode=NG1+1AA&sort_by=newestGET /v1/calculators/stamp-duty?price=300000&additional_property=truePOST /v1/property/reportwith{ "address": "10 Downing Street, SW1A 2AA" }
Full endpoint list in USER_GUIDE.md.
Environment Variables
Copy .env.example to .env. Key variables:
| Variable | Required For | Description |
|---|---|---|
EPC_API_EMAIL |
EPC lookups | Free key from EPC Register |
EPC_API_KEY |
EPC lookups | Paired with email above |
COMPANIES_HOUSE_API_KEY |
Company search | Free key from Companies House |
RIGHTMOVE_DELAY_SECONDS |
No (default 0.6s) | Rate limit delay for Rightmove scraping |
OPENAI_API_KEY |
Planning scraper | Vision-guided planning portal scraper |
Land Registry PPD and Rightmove work without credentials.
Development
# Install with dev extras
uv sync --extra dev
# Run API with reload
uv run uvicorn app.main:app --reload
# Run tests (mocked, no network)
uv run --extra dev pytest -v
# Run live integration tests (real network calls)
RUN_LIVE_TESTS=1 uv run --extra dev pytest -v
Architecture
Three-layer separation — core stays framework-agnostic:
property_core/ Pure Python library (all business logic)
app/ FastAPI wrapper (thin HTTP layer)
property_cli/ Typer CLI (thin CLI layer)
mcp_server/ FastMCP wrapper (thin MCP layer for AI hosts)
All three consumers import directly from property_core. No adapter layers.
Deploy (Fly.io)
fly secrets set EPC_API_EMAIL=... EPC_API_KEY=...
fly deploy
Deployed at https://property-shared.fly.dev with API docs at /docs and MCP endpoint at /mcp.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。