Hunt for exoplanet candidates in public telescope data with Claude Code
A Reddit post, widely discussed on HN, describes an amateur using Claude Code to search public data and find what he thinks is an unknown planet.
Real things people did with Researcher work, sourced and with recipes.
A Reddit post, widely discussed on HN, describes an amateur using Claude Code to search public data and find what he thinks is an unknown planet.
A researcher used a team of Claude Opus agents to run physics simulations and surface two candidate magnets for spin-based memory, then published every calculation.
A team ran about 17k coding-agent sessions to see which databases and services Claude Code, Codex and Cursor choose.
A developer built an R&D skill for Codex, Claude Code, Copilot, Gemini CLI and OpenCode with traceable artifacts and gates.
A developer ran Codex in an auto-research loop for a GPU Mode contest and reached a 232x speedup over baseline on QR decomposition.
An open-source terminal research agent enforces a spending cap, attaches a source to every claim and keeps local data local.
Open-source agent skills check any YouTube video, article, tweet or PDF claim by claim, with sources and a 0-10 score.
A builder runs up to 20 GPT-powered agent harnesses in parallel on open ErdΕs problems, checking proofs in Lean 4.
An open-source maintainer used Claude Code on a beach to show common geo and climate operations can be written as SQL, and even built a neural network in SQL.
A researcher with no web background built an AI tool for scientific figures with Claude and Cursor and reports about $1K MRR 25 days after launch.
Five real AI agent use cases each morning, each with a recipe and a source. Plus a Sunday roundup. Free.