graphify

A tool that turns a codebase (or docs, papers, transcripts) into a persistent knowledge graph: nodes for the code's real structures, communities for its neighborhoods, god nodes for its load-bearing centers. Queries run locally against the built graph and return focused subgraphs with source locations.

Every session used to start by rediscovering the codebase — expensive, slow, and different each time. Building the map once and querying it forever turns orientation from a per-session cost into a standing asset.

One full build per active project, then a post-commit hook keeps it fresh. Every session queries it before grepping — 'what calls this, how do these connect, where does this live' — for near-zero token cost. The graph this page renders is built by the same machinery, pointed at my working method instead of code.

The whole method is downloadable and adaptable — skills/graphify/SKILL.md inside the zip describes the workflow; the CLI builds and queries graphs from any repo.