Axiomind: Compiles Daily Notes into Axioms
Axiomind is a lightweight Markdown-based protocol that transforms unstructured daily notes into a structured personal cognition system using LLM agents. It organizes insights into categories like axioms, principles, projects, and areas, maintaining an auditable, Obsidian-compatible knowledge base. Unlike general knowledge wikis, Axiomind focuses on distilling personal beliefs and actionable rules from daily life inputs.
- ▪Axiomind is a protocol, not a note-taking app, designed to compile daily notes into a structured personal knowledge system.
- ▪It uses LLM agents to extract observations and propose axioms and principles, which require human review before activation.
- ▪The system outputs an Obsidian-compatible 'brain' directory with semantic MOCs, audit logs, and categorized knowledge assets.
- ▪Axiomind differs from Andrej Karpathy's llm-wiki by focusing specifically on personal cognition rather than general topic-based wikis.
- ▪The protocol emphasizes human oversight in promoting core beliefs and maintaining separation between noisy inputs and stable personal knowledge.
Opening excerpt (first ~120 words) tap to expand
Axiomind Axiomind is a lightweight protocol for turning messy daily notes into an agent-maintained personal cognition system. It is not a note-taking app, a RAG pipeline, or another folder template. It is a copy-pasteable Markdown protocol that tells an LLM coding agent how to compile your journals into an Obsidian-compatible brain/ made of observations, principles, axioms, areas, projects, resources, MOCs, and an audit log. Early demo: a real Axiomind brain has been successfully rendered in Obsidian Graph View: assets/axiomind-ob-graph-view.png. The core idea Most personal knowledge systems help you store notes. Axiomind helps you consolidate them. Daily notes are noisy: repeated templates, todos, mood logs, links, half-formed ideas, project updates, and occasional deep insights.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at GitHub.