richard-kim-79/archora-skills

This is a set of agent skills, reusable AI-powered commands, designed to assist with academic research when working with Claude Code or other compatible tools. The idea is to give researchers an AI assistant that can handle structured steps of the research process: generating testable hypotheses from notes, designing experiments, writing paper drafts in standard academic format (introduction, methods, results, discussion), checking for statistical errors, simulating peer review from multiple reviewer perspectives, and more. The skills are meant to chain together into a full pipeline. You start with raw notes or documents, organize them into a wiki, surface themes across them, form hypotheses, plan experiments, validate the logic and statistics, write the paper, then stress-test it with devil's advocate critiques and simulated journal review before submitting. A researcher or academic who works with Claude Code locally would install this to turn their AI assistant into a structured research partner rather than a general-purpose chatbot. It also connects to a web platform called Archora for team-based workflows. The project is MIT-licensed and has no primary programming language since the skills are instruction-based rather than code-based.