A spy satellite simulator just spent a few days as the number one trending project on GitHub. The globe spins in your browser, but the flights, ships, and satellites on it are real. I found it through a video by Matt Wolfe, who sat down with the mind behind it, Bilawal Sidhu, to dig into what it does, how he built it, and where the privacy line sits. It’s called God’s Eye View, it’s open source, and I couldn’t stop playing with it once I saw the demo.
What’s New
The creator describes it as “a spy satellite simulator in your browser, except the underlying data is real.” You know the command center dashboard from every Hollywood thriller, the one that shows every plane and vessel on Earth? That’s the vibe, except it pulls from open-source intelligence feeds instead of movie magic.
Here’s what you can layer onto the 3D globe:
- Live flights, with commercial traffic in white and military traffic in yellow. Click a plane and you get the callsign, registration, operator, origin, and destination. Military planes often show “operator unknown,” which is half the fun.
- Satellites currently in orbit, plus recent and upcoming rocket launches with a replay mode that shows the approximate trajectory and resulting orbit.
- Vessels on the ocean, each with unique identifiers.
- Traffic density from TomTom, color-coded like Google Maps, so red means jammed.
- Public CCTV cameras in select cities, showing where each camera points and what it currently sees.
- Infrastructure layers like data centers, dams, and submarine cables.
- Earthquakes from the past 24 hours and active fires detected by NASA’s FIRMS thermal satellites.
And you can talk to it. Say “take me to the busiest airport in the United States” and the camera flies to Atlanta. Say “start a camera orbit around the Texas capitol” and it does. There’s a cockpit view that puts you inside a plane about to land at SFO in real time, and a context mode that shows everything nearby.
The Twist
The part I didn’t expect is that this isn’t just a toy. Sidhu showed how people use it to reconstruct real-world events. Someone took satellite provider data and rebuilt the Iran flooding, plotting the flood’s path and lining up user-generated clips along that trajectory so you can watch what happened in sequence instead of doom scrolling through random videos. He also rebuilt a plane crash using NTSB drone footage, ATC audio, a 3D scan, and ground-level imagery, all lined up spatially so you can see the skid marks on the runway.
His framing stuck with me: right now, when something happens, you get either social media chaos or “one dot on a map with four people yapping.” He wants a middle option. Civilian intelligence that gets you as close to the ground truth as possible.
The other twist is how it got built. The expert said the first version came together in March and April using OpenClaw driving Gemini 3.1 and Opus, with Gemini handling spatial reasoning and Opus doing the full-stack work. He described it as a “fever dream for a couple of days”: foundation first, then satellites, then planes, one layer at a time.
Privacy: Where the Hard Line Is
Matt asked the obvious question. Couldn’t someone use this to stalk me? The creator’s answer was clear: the open-source project has a hard rule that you cannot track people. It’s about large-scale infrastructure, not individuals.
A few specifics worth knowing:
- CCTV feeds are delayed. In Austin you get roughly one frame every 10 minutes, on purpose.
- Traffic data is aggregated and anonymized. No license plates, no following a car through town.
- Flight tracking uses the same public data you can already find with a Google search.
- Live layers run one step behind real time to keep the animation smooth.
Sidhu’s argument is that surveillance tools already exist and governments use them constantly. This is the opposite: a window that lets regular people see the same public data in one place.
Step-by-Step Mini-Workflow
Matt showed the laziest possible install, and I love it. You don’t touch a terminal.
- 🛰️ Grab the GitHub repo URL for God’s Eye View (link in the video description).
- 📁 Open the ChatGPT desktop app or Claude Code app, create a new local project, and point it at an empty folder.
- 🤖 Paste this prompt: “Clone this GitHub repo. Read the instructions. Install all necessary dependencies, and then tell me what I need to do to finalize the setup and get everything working.” Add the repo link and submit.
- ⏱️ Wait about two minutes. The agent clones, installs, and opens the app in your browser.
- 🔑 Open the “Power Up” section in the app. Each data layer has a “get key” link that takes you straight to the sign-up page for that API. Paste your keys and the layers switch on one by one.
Pro Tips
- Budget check. The creator kept the open-source version as cheap as possible. For personal, non-commercial use, most people pay nothing. Heavy daily use might run a few dollars a month. The one exception is voice interaction, which uses a real-time ChatGPT API and can get pricey fast. Skip it if you’re cost-sensitive.
- Ask the agent “why.” Find a cluster of military flights and ask what’s going on there. The model’s world knowledge usually explains it: a base, helicopter training, that sort of thing. It’s a fun way to learn geography without meaning to.
- Use it as ambient TV. Sidhu keeps his favorite airport open while coding. I think that beats a lo-fi stream.
- Content creators, take note. People are already using it to generate spy-scene footage for videos. Camera orbits plus real data look expensive on screen.
- Treat the rocket replays as approximate. The creator said outright that he doesn’t have the physics degree or the compute to model exact trajectories.
The full video walks through every layer live, including the crash reconstruction and the San Diego military traffic demo. Go watch it, then spend an evening spinning the globe 🌍