Tools · Personal ops

Ogden Home Finder

Every morning an AI agent scrapes listing sites for housing in the Ogden metro, and a gate checks its work before anything ships. This page is the living result — price history, days on market, and a map. Flip between rentals and for-sale homes.

—Active
—Missing
—Removed
—Avg rent
—Lowest
—Last run

How it works: every morning a GitHub Actions job has a Claude agent run two scrapers over everything within about ten miles of Ogden, from Pleasant View down to Layton and out to Hooper and Huntsville: rentals from rent.com’s listing data, for-sale homes from Redfin’s. Each pin is where the listing site puts the home. A merge script folds each run into a cumulative JSON catalogue that tracks first-seen date, price changes and disappearances. “Active” means a scrape saw the listing in the last week; one unseen for seven days drops to missing. A validation gate refuses any run that would delete listings, rewrite history or add duplicates. Two known gaps: some homes appear twice (once per listing site) until a dedupe pass lands, and listings no scrape has seen since pins started coming from the source have no real location, so they stay in the table but off the map.How the pipeline works. This page fetches the catalogues at load time and renders everything client-side. Nothing is sent anywhere.

This is what personal automation looks like. A scheduled agent, a static page, a JSON file, and a gate that checks the agent’s work before it ships. No backend, no database, no login.

Email meHow it’s builtMore tools