YOU

Things you didn’t know you could do with Claude. Part 37! Full prompt below - just make sure you switch in your specific criteria! Role. You're my property investigator in Chrome/Cowork. I paste one listing URL; you open it, pull real data from public sources, and build me a single interactive visual dashboard — not text cards. The angle is "what the property sites won't tell you," so every panel is something the listing hides, shown as a visual I can read in two seconds. Input. One property URL (Rightmove / Zoopla / OnTheMarket). Step 1 — Read & capture. Open the listing, screenshot the main photo, and pull address, postcode, lat/long, price, beds, EPC, tenure, lease, service charge, price history, and days on market. Step 2 — Pull real data and render each as its visual. For each, hit the live source, cite it, and grade confidence (Confirmed / Partial / Couldn't verify — never invent): Crime → sparkline + category donut. Use the police.uk open API for street-level crime around the postcode. Show a 12-month trend sparkline with an up/down arrow, and a breakdown of the top categories. Label it "area, not address." Flood → map overlay. Use Environment Agency flood-risk data. Render the flood zone on a map with the house pinned, plus any past-flooding notes from local news. Sold prices → line chart. Use HM Land Registry Price Paid for the street/postcode. Plot the trend and mark this house's own last sale; make the asking-vs-reality gap visible. Location → plotted map with distances. OpenStreetMap: pin the house, nearest water, and nearest mainline station, with real walking/driving times drawn as rings. This is where "near water" and "30 min to rail" get proven or busted. EPC → gauge. Pull the actual certificate from the EPC Open Data register; show the rating against the local average. Planning → timeline. Council planning portal: any applications on or near the property, as a dated list with the one that matters flagged. Listing tells → annotations. Rooms with no photo, "cosy," reductions, long time on market — call them out. Step 3 — Assemble the dashboard. One interactive HTML page: screenshot and headline facts up top, then a fit-score gauge out of 10 (against 3 beds / water / ≤30 min rail / £450k), then each check as its own live visual panel with a confidence badge and linked source. Any red flag (flood history, rising violent crime, view-blocking planning, short lease) surfaces at the very top. End with a one-line "here's the catch." Rules. Real data only, one source per panel, confidence graded honestly. Make it genuinely interactive — hover for detail, filter where it helps. I want a dashboard I'd screenshot, not a report I'd skim. #claude #ai #propertysearch #property #claudeai