The part that actually matters
Most idea trackers are glorified spreadsheets: a list of half-formed thoughts nobody ever revisits. Mine started that way too, until the part that actually mattered turned out not to be capturing ideas, but killing them properly.
The pipeline
Startup-Analyser is an open-source, depersonalised fork of a private tool I built for myself: a markdown-and-git pipeline for capturing, scoring and killing venture ideas.
The scoring model is deliberately narrow. Every idea gets rated against six founder-fit constraints and five market dimensions, which places it on a 2x2: Build, Opportunity Gap, Lifestyle or Kill. No SaaS dashboard, no database. Markdown files, git history, and a CLAUDE.md that tells an agent how to run the whole pipeline on your behalf.
The kill
Most ideas that get scored end up in Kill. That's fine, and it's the point. The bet behind the design is that a killed idea is only wasted if you let it be. Every kill ceremony forces one more step: name the pattern in the idea that failed and write it into patterns.md. Over enough cycles that file becomes more valuable than any single idea in it, a running record of the mistakes in your own thinking, made explicit enough that you stop making them twice.
Why it's forked
Startup-Analyser strips out everything specific to me: my own scoring history, my personal context, and the provenance of my private knowledge base. What's left is the reusable mechanism, for anyone who wants to run the same discipline against their own ideas.
Stack
Markdown, git, Claude Code (CLAUDE.md-driven pipeline).