Claude Code Helps Independent Developer Spot Unseen Planet in Public NASA Telescope Records
A hobbyist sitting at home with an internet connection and smart computer programs may have just found a brand new world that professional astronomers missed. By feeding raw space data into modern artificial intelligence tools, including Anthropic's coding assistant Claude Code, an independent developer discovered telltale signs of an unmapped exoplanet orbiting a distant star. The intriguing discovery shows how cutting-edge artificial intelligence agents are allowing curious everyday citizens to tackle complex scientific problems that once belonged only to university research teams.
The search for planets outside our solar system typically requires teams of astronomers to analyze mountains of light measurements captured by space observatories. NASA makes these massive archives available to the public, but the datasets are so huge that millions of subtle clues remain buried under endless lines of numbers. When a planet passes directly between its host star and an observing telescope, it briefly blocks a tiny fraction of the star's light. By looking for rhythmic dips in brightness that repeat like clockwork, planet hunters can identify worlds traveling hundreds of light years away.
The breakthrough emerged on Wednesday, October 7, 2026, when developer Pavel Rabtsevich shared his findings online. Rabtsevich, a technology worker living in Spain with no formal background in professional astronomy or access to a physical observatory, decided to explore archival records collected by NASA's Transiting Exoplanet Survey Satellite, known as TESS. While scanning through the records, he noticed a faint star roughly 116 light-years away from Earth that appeared to drop in brightness for about two hours every 3.18 days.
Spotting a dimming pattern in raw data is exciting, but it is only the first step. Many non-planetary events can fool observers, including background twin stars orbiting each other, camera sensor glitches, cosmic ray strikes, or ordinary magnetic activity on the star's surface. To test whether the dip was genuine, Rabtsevich needed to run dozens of rigorous scientific checks. Instead of writing all the specialized mathematical software from scratch, he paired Claude Code with other advanced artificial intelligence systems to build and execute an automated testing pipeline.
The artificial intelligence assistant wrote custom code to clean the noisy data, correct for spacecraft movement, and model the exact shape of the light curve. Rabtsevich instructed the system to run seventy-four distinct validation tests to rule out false positives. The tools even successfully predicted a previously hidden dimming event in older data from 2018 and 2020 that had been withheld from the initial algorithm, proving the mathematical pattern held steady across nearly a decade of observations.
Astronomers remain cautious and note that the potential world is currently classified as a candidate planet rather than an official discovery. Nature can sometimes create rare optical illusions that mimic the transit of a planet. To verify whether the object is truly a solid world, professional scientists must point ground-based observatories at the star to measure its physical mass using the slight gravitational wobble it causes on its parent sun.
Following the submission of the research paper and detailed test logs, NASA scheduled official ground-based observation sessions to verify the candidate. If the follow-up measurements confirm the signal, the discovery will mark one of the clearest examples yet of a non-expert uncovering a new planetary body through conversational coding tools. The object is estimated to be a warm world orbiting close to its host star, completing an entire year in just over three Earth days.
The project highlights a profound shift in modern science, turning public datasets into fertile ground for amateur researchers. As artificial intelligence models become better at writing accurate software, designing laboratory tests, and spotting subtle mathematical trends, the wall between academic institutions and home enthusiasts continues to crumble. Whether the candidate proves to be a rocky planet or an unusual stellar twin, the experiment proves that the future of cosmic discovery may happen right from an ordinary home desk.