AI Native Landscape

graphify

Tracked

A skill that turns any codebase — with its docs, SQL schemas, configs, and PDFs — into a queryable knowledge graph for Claude Code, Cursor, Codex, and Gemini CLI. Local deterministic AST parsing, every edge explained, no vector store.

AuthorGraphify-LabsOpen Sourced2026-04-03Last CommitUnknown

Overview

Graphify maps an entire project (code, docs, PDFs, images, videos) into a knowledge graph you query instead of grepping through files. Type /graphify in your AI coding assistant and it produces an interactive graph.html, a highlights report, and a graph.json you can query anytime without re-reading the files. It is deliberately not a vector index — there are no embeddings, just a real graph you traverse to ask questions, trace paths, and explain concepts.

Key Features

  • Code, docs, SQL schemas, configs, and PDFs mapped into one queryable knowledge graph
  • Code parsed locally with tree-sitter AST — deterministic, no LLM, nothing leaves your machine
  • Every edge tagged EXTRACTED (explicit in the source) or INFERRED (resolved by graphify), so you can tell what was read directly from what was inferred
  • God nodes surface the most-connected concepts; Leiden communities split the graph into subsystems
  • Query commands: plain-language questions, shortest path between any two things, and per-concept explanations
  • Works with Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot, and 20+ other assistants on macOS, Linux, and Windows

Use Cases

  • Onboard onto or explore a large unfamiliar codebase by querying the graph instead of reading files
  • Trace how two distant parts of a system connect, hop by hop with per-edge confidence
  • Keep a persistent graph.json that agents query without re-reading the repository

Technical Details

  • Code is parsed with tree-sitter AST covering ~40 languages, resolving cross-file calls, imports, inherits, and mixes_in edges deterministically; docs, PDFs, images, and video get a semantic pass using the assistant's own model or a configured API key
  • Every edge carries a confidence tag (EXTRACTED vs INFERRED); graphify path A B reports the shortest path with the overall confidence of its weakest hop, and graphify query returns a scoped subgraph for a natural-language question
  • Leiden community detection clusters the graph into subsystems with LLM-free labels; no embeddings or vector store are used — graph.json is traversed directly and carries schema_version and graphify_version fields for format compatibility
  • Installed as the graphifyy PyPI package (Python 3.10+, via uv or pipx); graphify install auto-detects the assistant and registers the skill