Deploy and Host property-radar on Railway

Property Radar is a self-hosted MCP server for discovering and monitoring properties for sale or rent. It crawls listings into your own Postgres database and remembers how you reacted to each one, so Claude or ChatGPT can pick your home search back up in a new conversation instead of starting from zero. The server never calls a model itself; the LLM you connect does the judging.

About Hosting property-radar

Property Radar is five services that this template provisions together: mcpd (one URL serving MCP, a REST API, interactive docs, and a /connect setup page), an always-on crawler worker, a small web console, Postgres for listings and taste state, and versitygw, an S3-compatible thumbnail cache on a volume. You type zero secrets: the MCP bearer token, the image-proxy key, and the storage key are generated per deploy. After the build, open /connect on the mcpd domain and paste the one claude mcp add line it shows. The only optional input is a residential Webshare proxy key on the crawler; without it the cloud crawler idles and you can crawl from your own machine into the same database. Typical cost is a few dollars a month.

Common Use Cases

Give Claude or ChatGPT a persistent memory of your apartment or house hunt: verdicts, a taste rubric, and favorites lists that survive across conversations
Ask a fresh conversation for "3–5 new homes I'd like" and get picks with your own past reactions as the reasons, judged from the actual listing photos
Keep a private, always-on listings corpus for a metro (NYC via StreetEasy today) with price history and lifecycle tracking, reachable from hosted MCP clients like claude.ai

Dependencies for property-radar Hosting

Postgres (provisioned by the template)
An S3-compatible object store for thumbnails (versitygw on a Railway volume, provisioned by the template)
Optional: a residential Webshare proxy plan, for crawling from Railway's datacenter IPs

Implementation Details

Connect a client after deploy:

claude mcp add property-radar --transport http https:// \
  --header "Authorization: Bearer "

URL-only connectors (claude.ai, ChatGPT) pass the token as ?token=…. Interactive REST docs live at /docs, health at /healthz. Set SEED_CRAWL_AREAS on the crawler to pre-seed a standing crawl scope, or let the model queue one with request_crawl. Source, docs, and the full tool list: https://github.com/davidteather/property-radar

Why Deploy property-radar on Railway?

Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it.

By deploying property-radar on Railway, you are one step closer to supporting a complete full-stack application with minimal burden. Host your servers, databases, AI agents, and more on Railway.