Claude Code helped spot a potential exoplanet — TIC 4206066 b, a candidate world 1.4 times Earth’s size orbiting its star every three days. Pavel Rabtsevich, a space enthusiast and AI user in Belarus, noticed the periodic dimming in NASA’s public TESS data, then used Claude Code to model, test, and validate the transit signal across 74 separate runs. The AI not only confirmed consistency in the light-curve pattern but predicted exact dimming windows, even when fed incomplete data. Rabtsevich submitted the finding to the TESS Science Office, which accepted it for follow-up observation. The telescope is scheduled to reobserve that patch of sky in November 2026. If the observed dimming matches Rabtsevich’s predictions down to timing and depth, formal confirmation by NASA could follow. No peer-reviewed paper has been published. No observatory has yet verified the signal independently. And no planet name has been assigned. Only the provisional designation tied to its host star, TIC 4206066. This isn’t the first time an AI-assisted amateur discovery has entered NASA’s pipeline, but it’s among the first where Claude Code, not a custom-trained model or academic pipeline, was the primary analytical tool. It’s also the first time Claude Code has been cited in a TESS follow-up request. The story isn’t about coding. It’s about access: public data, open telemetry, and an AI tool that lets one person run planetary detection logic without writing a single line of Python. Rabtsevich didn’t build a telescope. He used what was already online. And asked Claude Code to help him read it right.