Skip to main content

What Is a Knowledge Graph

A knowledge graph is a structured collection of interconnected facts — entities and the relationships between them. In KMGraph, the "entities" are your lessons, decisions, and patterns; the "relationships" are the links between them.

Why graph, not document?​

A flat document (like a wiki page) captures information but doesn't capture connections. A knowledge graph captures both. When you link an ADR to the lesson that motivated it, the graph knows that context — and so does your AI.

How KMGraph implements it​

KMGraph's knowledge graph is a directory of Markdown files with structured frontmatter. Each file is a node. Links between files are edges. The FTS5 search index and MEMORY.md index make the graph traversable without a graph database.

This means:

  • Entries are plain text — readable without KMGraph
  • The graph lives in git — versionable, portable, team-shareable
  • No server or database required