6 days ago
411
10
Posted On: October 11th 2026, 07:04 pm
Someone used Claude Code to identify a potential new planet about 116 light-years from Earth, and NASA’s TESS mission has approved follow-up observations to test it.
Independent researcher Pavel Rabtsevich used Claude Code with Opus 5.5 to analyze years of TESS data. He found a star dimming every 3.18 days, with the same pattern appearing across observations from 2018, 2020 and 2025. If the signal is caused by a planet, the candidate would be roughly 1.4× Earth’s radius.
Over two weeks, Rabtsevich ran 74 analyses and generated more than 1,000 scripts with Claude Code and Codex to test the signal and rule out other explanations. NASA’s TESS mission has now approved new observations, and Rabtsevich published the exact times he expects the star to dim in November.
The planet hasn’t been confirmed yet. If the signal appears again on schedule, the case gets much stronger.
Comment “SEND” and we’ll send you the interactive 3D system. 👇
Follow @longliveai for more.
#ai #claude #nasa #space #astronomy
Independent researcher Pavel Rabtsevich used Claude Code with Opus 5.5 to analyze years of TESS data. He found a star dimming every 3.18 days, with the same pattern appearing across observations from 2018, 2020 and 2025. If the signal is caused by a planet, the candidate would be roughly 1.4× Earth’s radius.
Over two weeks, Rabtsevich ran 74 analyses and generated more than 1,000 scripts with Claude Code and Codex to test the signal and rule out other explanations. NASA’s TESS mission has now approved new observations, and Rabtsevich published the exact times he expects the star to dim in November.
The planet hasn’t been confirmed yet. If the signal appears again on schedule, the case gets much stronger.
Comment “SEND” and we’ll send you the interactive 3D system. 👇
Follow @longliveai for more.
#ai #claude #nasa #space #astronomy
Comments (10)
5 days agomax_will_win
Wild part: the pattern sat in public TESS data for years. The model didn't get smarter data, it just looked harder. Now the follow-up observations decide if it's real. That's the right order: model finds, telescopes confirm.
6 days agothegrokpack
Nice real-world use of Claude Code. Tip for anyone trying data analysis like this: ask the model to list its assumptions and what would falsify the result before you trust any finding.
More posts
