Your File System Is Already a Graph Database

A personal knowledge base built from markdown files and wikilinks in Obsidian can function as a graph database with LLMs as the query engine. Files are nodes, wikilinks are edges, and folder taxonomy acts as schema — no vector store or RAG pipeline needed. The author, who maintains 52,000+ files, describes a PARA-inspired folder structure and a daily workflow where an AI agent creates meeting notes, links them to people and projects, and aggregates context from Slack, Google Docs, and other sources. The key insight is that this isn't just a wiki — it's a context engineering system that gives LLMs the real project history needed to produce high-quality outputs like design docs and handoffs. The unsolved challenge is automated inbox processing that stays consistent over time.