Graphify: turn any codebase into a knowledge graph

TL;DR: Graphify turns a repository, its documentation, SQL schemas, config files and PDFs into a knowledge graph you can query. It runs as a /graphify skill inside Claude Code, Cursor, Codex and Gemini CLI, and it builds that graph with local deterministic AST parsing rather than a vector store.

What it is

Graphify is a code intelligence tool built for AI coding agents. The repository description is specific about the scope: it turns any codebase, together with its docs, SQL schemas, configs and PDFs, into a queryable knowledge graph. That is a wider input surface than most code graph tools handle, since they usually stop at source files in a single language.

The distribution path is a skill, not a library you wire into your own agent harness. The description names Claude Code, Cursor, Codex and Gemini CLI explicitly, and the repository topics confirm the same set of integrations, including antigravity and claude-code.

The repository sits at 131,706 stars with 12,641 forks and 1,612 open issues. It is owned by an organization, Graphify Labs, and it was created on 2026-04-03. The default branch is v8, which tells you the project iterates on a numbered release line rather than tracking main.

How the graph gets built

Three properties in the description shape how the tool behaves:

• Local deterministic AST parsing. The graph comes from parsing source directly, so the same input produces the same graph on any machine.
• Every edge explained. Each relationship carries the reason it exists instead of a similarity score pulled out of a vector space.
• No vector store. There is no embedding database to run, index or pay for.

That last point separates Graphify from the usual pile of code search tools. Deterministic parsing gives you a graph you can diff across commits and a story you can argue about in review.

The source surface it accepts is wider than code:

Input	What Graphify says it handles
Source code	Parsed with deterministic AST analysis
Documentation	Included in the graph
SQL schemas	Included in the graph
Config files	Included in the graph
PDFs	Included in the graph

How to try it

The repository is Python and the install path is the skill, so start by reading the code and the skill definition:

git clone --branch v8 https://github.com/Graphify-Labs/graphify.git
cd graphify
ls

In your agent, the entry point is a single slash command:

# Claude Code, Cursor, Codex or Gemini CLI
/graphify

The skill then walks the project, parses it, and writes out the graph.

When to use it

Graphify makes sense when your agent keeps re-reading the same files because it has no durable map of the codebase, and when your questions are structural rather than “grep for X”. A graph with explained edges also helps when a question crosses module boundaries, since you can follow the relationship instead of guessing which file to open.

When to skip it

Two cases. If your repository is a single small service with a handful of files, a plain file listing is faster and a graph adds nothing. And if you specifically want semantic similarity search over prose, a graph is the wrong shape, because the description says outright that there is no vector store.

The project is young, created in April 2026, and carries 1,612 open issues against 131,706 stars. That ratio is normal for a fast-growing open source project, but it does mean you should expect rough edges.

Source and Accuracy Notes

• Project page: https://github.com/Graphify-Labs/graphify
• Source repository: Graphify-Labs/graphify (verified via GitHub API html_url)
• License: Apache-2.0 (verified via GitHub API license.spdx_id)
• Stars at time of writing: 131706