AI-Powered Academic Research From Idea to Published Paper
github.com/regisely/superpapersRaw chat is useful, but serious research needs a structured way to think with AI.
AI can suggest many paths, but there is no disciplined way to debate trade-offs and converge on one design.
Good ideas vanish across prompts when assumptions, decisions, and rejected alternatives are not documented.
Without a structured pipeline, AI-assisted research turns into scattered chat instead of a traceable paper workflow.
Superpapers turns AI into a research partner.
Superpapers gives researchers a structured way to think with Claude, converge on a design, and keep the full workflow documented and reproducible.
Brainstorm. Plan. Execute. Submit.
Each stage leaves artifacts behind: design choices, execution plans, scripts, outputs, and a reproducible paper trail.
The plugin challenges question, design, data, and contribution before execution.
Tables and figures stay tied to scripts and raw data.
Search, data, modeling, writing, and submission in one system.
Independent tasks run in parallel under one reviewed plan.
Built for any empirical quantitative workflow.
Idea to manuscript, without handoff gaps.
24 screenshots from a real Claude Code session that produced the paper available on the repository page.
Brainstorming, design, estimation, robustness, writing, and submission formatting happen in the same research thread.
The first identification strategy looked promising, then failed after estimation. The workflow changes route instead of forcing a weak result.
Failed first stages, null results, robustness checks, and reframing decisions remain explicit in the record.
Claude proposes options, but major choices are discussed with the user before the project moves forward.
24 screenshots from brainstorm to submission. Arrows step through.
Act I · Brainstorm
Add the GitHub marketplace, install the plugin, and start talking about your research.
/plugin marketplace add regisely/superpapers
/plugin install superpapers