Graphify-Labs/graphify

What it does

A skill for Claude Code, Codex, OpenCode, Cursor, Gemini CLI and more: type /graphify and your entire project — code, SQL schemas, docs, PDFs, images, videos — is mapped into a queryable knowledge graph, replacing grep-and-scroll. Code parsing uses tree-sitter AST: deterministic, no LLM, fully local; only document content goes through a model. Not a vector index — a real graph, with every edge labeled EXTRACTED (read directly) or INFERRED.

What's inside

Distributed as a CLI (PyPI package graphifyy, install via uv tool install or pipx); the repo holds docs/ (README translations in 30+ languages) and interactive graph.html visualizations; default branch v8, iterating fast.

Tech stack

Python + tree-sitter, outputting force-directed HTML graphs; hooks into each CLI assistant as a skill.

Use cases

• Developers inheriting large legacy projects — one command maps the codebase into a knowledge graph
• Heavy AI users — swap vector RAG for graph context that retrieves the right thing
• Docs teams — fold docs, PDFs and diagrams into one queryable graph

Why it's trending

Its knowledge-graph alternative to vector-only retrieval speaks directly to limitations teams encounter when building contextual AI assistants.

Current chart placements