A machine-readable graph of truth claims, built on Git and Markdown

Name: A machine-readable graph of truth claims, built on Git and Markdown
Availability: InStock
Author: maximilliann

by maximilliann·Feb 28, 2026·1 point·0 comments

Visit Project View on HN

AI Analysis

●●●BangerZero to OneBig BrainRabbit Hole

Philosophy-as-code: Git-backed argument DAGs parseable by both humans and LLMs.

Strengths

•Genuinely novel infrastructure: treats logical arguments as dependency graphs (like code), enabling version control and cross-repo citations.
•LLM-aware design: breaking claims into discrete subclaims plays to LLM strengths instead of fighting their limitations.
•Minimal, elegant constraint: Markdown + Git solves distribution, versioning, and attribution without a database lock-in.

Weaknesses

•Tiny immediate audience—only appeals to philosophers and AI researchers, not a general tool.
•No live ecosystem yet: the demo repo is a single Descartes argument tree. Network effects require adoption.

Post Description

I've long been fascinated with deductive reasoning and wondered if it is possible to create comprehensive maps of human knowledge, starting from first principles.

At it's core, Prime is just tooling to write, publish and explore a decentralized DAG of truth claims hosted on Github; a kind of infrastructure for rational thinking at scale.

With the advent of LLMs I think this project might be more worthwhile than ever. I find LLMs can actually be quite "rational" if given specific tasks, but tend to get lost in language (just like humans). I'm hypothesising that breaking arguments down into discrete subclaims plays nice with LLMs, unlocking new ways of working in fields such as philosophy, science and political discourse.

Hope you find it interesting!