Selected Chronology

A curated selection of my recent writings. I hope you find something useful out of these thoughts and observations.

How I built a knowledge graph of 0 A.D.'s entire codebase and slashed my AI token costs by 70x

AI coding agents waste huge amounts of time and tokens reading raw files from scratch every session. This article walks through building a persistent knowledge graph of the 0 A.D. codebase using graphify - an open source tool that combines a free local AST pass with an optional LLM powered semantic pass. The result: a queryable graph that slashes token costs by up to 70x, persists across sessions, and integrates with VS Code via MCP for agentic coding.

Beyond tables & vectors: why your most complex problems need a knowledge graph

When relational databases buckle under recursive joins and vector databases can only guess at similarity, knowledge graphs excel at tracing deterministic, multi-step relationships. This article uses a real-world software dependency errata handling scenario to show how Stardog's SPARQL queries effortlessly answer "who is transitively affected?" — and explores broader use cases like end-to-end requirements traceability and grounding AI coding agents in a precise, queryable map of your codebase.