Just launched alive-analysis — traceable AI analysis workflows (Claude Code & Cursor, open source)
Hey everyone,
I just launched alive-analysis — an open-source workflow kit that runs inside Claude Code and Cursor.
Problem: you ask AI to analyze something, get a good answer, then a month later nobody can explain why we concluded what we did. Chat is gone and the logic isn’t reproducible.
alive-analysis fixes that by turning the agent into a structured analysis partner. Instead of one-shot answers, it guides you through a 5-step loop (Ask → Look → Investigate → Voice → Evolve) and saves each analysis as Markdown files you can Git-track and search later. Checklists nudge you to consider confounders, Simpson’s paradox, and counter-metrics. Two modes: Quick (single file) and Full (multi-file + quality gates). Free and open source.
It’s scheduled to go live on the homepage on Feb 19. If you do data work with Claude Code or Cursor, I’d love feedback on the checklists or how you’d want it to fit into your repo.
GitHub: https://github.com/with-geun/alive-analysis




Replies