graphify: an AI coding assistant skill that turns codebases into queryable knowledge graphs

graphify is a universal skill for 20+ AI coding assistants (Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, etc.). After installation, the /graphify command maps an entire project into a knowledge graph. Code is parsed locally with tree-sitter ASTs — no LLM calls, no data leaving the machine — resolving cross-file calls, imports, and inheritance across ~40 languages. Docs, PDFs, images, and videos are handled semantically via the assistant's own model or a user-configured API key. Every edge is labeled EXTRACTED (explicit in source) or INFERRED (derived by parsing), and output includes graph.html, GRAPH_REPORT.md, and graph.json. Installation requires Python 3.10+ with uv or pipx; two commands complete CLI install and skill registration. Once built, developers query the graph via explain, path, and query commands instead of reading files — e.g. tracing a 3-hop path from APIRouter to ModelField in a FastAPI codebase. Benchmarks show recall@10 of 0.497 on LOCOMO versus mem0's 0.048, with zero LLM credits consumed during graph construction. For teams maintaining large monorepos or needing cross-repo code understanding, Graphify Cloud offers an always-on graph service claiming 80% lower agent token usage, with connectors for Sentry, Jira, and more. The local version is fully free, suited to developers who need code privacy and won't let source leave their machines.