chensitao

You can discover and ingest papers into a structured wiki, generate systematic literature reviews with relevance scoring, design and run experiments with verdict gates, draft and compile LaTeX papers, verify novelty, build Stata empirical plans, and produce rebuttals to reviewer comments.

Academic researchers, PhD students, and ML scientists preparing submissions to venues like ICLR, NeurIPS, ICML, ACL, and CVPR, plus empirical economics and management researchers who need Stata execution plans, variable mapping, and identification-strategy design from ingested literature.

Run the setup skill for interactive API key configuration covering Semantic Scholar, DeepXiv, and Review LLM via .env, then use init to bootstrap the wiki from your sources, ingest papers by PDF, .tex, or arXiv URL, and invoke the research orchestrator with a direction or brief.

The skills require API keys for Semantic Scholar, DeepXiv, and a Review LLM, configured through the interactive setup skill. LaTeX tooling such as latexmk is needed for paper compilation, and Stata is required to execute generated do-file skeletons for empirical work.

Yes. The research orchestrator supports session-resumable state with human gates, letting you restart from specific stages using --start-from flags. The reset skill restores a clean wiki scaffold by scope, and exp-status tracks running experiments with optional auto-collection and pipeline advancement.