Open-source tool transforms code, docs, and configs into a knowledge graph for AI assistants, using local AST parsing with no LLM calls or vector stores.

Graphify, an open-source codebase analysis tool, has reached 131,811 stars on GitHub with 12,649 forks since its April 2026 launch. The tool transforms any codebase—including documentation, PDFs, SQL schemas, and configuration files—into a queryable knowledge graph.

What it does

Graphify parses code locally using tree-sitter AST, producing deterministic results without calling an LLM or sending data off your machine. Docs, PDFs, images, and video are processed semantically using your assistant’s model or a configured API key.

The output is three files: graph.html (an interactive visualization in any browser), GRAPH_REPORT.md (key concepts and surprising connections), and graph.json (the full graph for persistent queries).

How it differs from vector search

Graphify is not a vector index. Instead of embeddings and similarity search, it builds a real graph where every edge is tagged as either EXTRACTED (explicit in the source) or INFERRED (resolved by graphify). This distinction lets developers know what was read directly versus what was derived. The graph supports queries like “explain APIRouter” or “trace the path between concept A and concept B,” and resolves cross-file links across roughly 40 languages via tree-sitter.

Getting started

Installation takes 30 seconds on Python 3.10+ with uv or pipx:

uv tool install graphifyy

Register with your assistant:

graphify install

Then in your AI assistant, run:

/graphify .

The tool works on macOS, Linux, and Windows with Claude Code, Cursor, Codex, Gemini CLI, and 20+ other assistants. It is licensed under Apache-2.0.

Local-first and cloud options

The open-source version requires no backend calls for code analysis. The README mentions Graphify Cloud, a paid service, which keeps the graph live across entire SDLCs, integrates with Sentry and Jira, and reports cutting agent token spend by 80%. The open-source version is fully local and deterministic.

Go deeper

Get the latest AI updates in the Tech AI Magazine app
iPhone · Android