graphify

A structural code-navigation layer that builds an AST-based index of a codebase — classes, functions, imports, and a call graph — using tree-sitter, so an agent can know what exists, where it lives, and how it connects before resorting to keyword search. It is delivered as a CLI (installed with npm i -g graphify-ts, which runs on the Bun runtime) and triggered via /graphify. The index is updated manually rather than automatically, a deliberate choice to avoid write conflicts in git worktrees and multi-session setups. Setup builds the index once per project with graphify build ., which scans all source files, extracts AST structure, and saves it to graphify-out/graph.json, reporting the counts of files, symbols, and relationships. Four commands drive it: build for the first-time or full rebuild; query, a case-insensitive search that returns matching symbols with their file locations; update, an incremental re-extraction of named files after edits; and auto-update, which computes changed and untracked files from git diff and refreshes them in bulk, staying silent when nothing changed. The recommended usage is to query the graph before Glob or Grep whenever graphify-out/graph.json exists, replacing blind keyword search with structured lookup; to run an incremental update after editing if the graph will be queried again in the same session; and to query concepts when exploring unfamiliar code to find entry points without guessing filenames. It supports a broad set of languages including Python, JavaScript, TypeScript (JSX/TSX), Go, Rust, Java, C, C++, Ruby, C#, Kotlin, Scala, and PHP. The graph output is JSON with nodes, edges, and metadata, and edge relations include contains, method, imports, imports_from, calls, and inherits.