Fusion Paper Scout
The Paper Scout method pointed at nuclear fusion. Corpus built — 2,600 records down to 51 papers — and a first experiment brief written.
What It Does
After reading Project Hail Mary I was thinking about efficient energy sources in reality without biologically evolved mass/energy conversion and I spent some time reading about present-day efficiency of nuclear fusion technology. I'm not a physicist but I read enough to wonder what research had been done on creating a more favorable quantum environment for fusion without just throwing more heat and pressure at it. So this applies my alt-protein Paper Scout project to that topic instead.
How it works
This was just a generalization of the same concept although the source of literature and comparative approach varied slightly. Fusion drops Europe PMC for arXiv/INSPIRE/OSTI/Semantic Scholar/ADS.
Built with
- Built with: Python 3.11 CLI stages (uv run, PEP-723 inline deps, httpx and stdlib only) running a search → keyword-filter → LLM-score → coverage-first-select pipeline over six academic APIs — OpenAlex, arXiv, INSPIRE, OSTI, Semantic Scholar, NASA ADS — with per-domain adapters that treat only OpenAlex as authoritative for absence, plus preflight probes and budget detection so a rate-limit never reads as an empty result.
- Output is a hand-authored print-styled HTML brief rendered to a single-scroll PDF via headless Chrome; no server, no database — JSON and CSV on disk, deliberately duplicated per domain rather than abstracted into a shared core.