How to Set Up Graphify Locally: Complete Installation Guide

Install Python 3.10+, run uv tool install graphifyy, then execute graphify install to register the skill and graphify . to build your first knowledge graph.

Setting up Graphify on your local machine equips you with a self-contained knowledge graph generator for any codebase. This guide covers the complete installation process for the Graphify-Labs/graphify repository, from CLI setup to AI assistant integration.

Prerequisites

Before installing Graphify, ensure your system meets these requirements:

• Python 3.10 or newer – Required for running the Tree-sitter AST parser and CLI tools
• uv (recommended) or pipx – Isolated environment managers that prevent dependency conflicts

Install the prerequisites using your platform's package manager:

# macOS (Homebrew)

brew install python@3.12 uv

# Windows (PowerShell)

winget install astral-sh.uv

# Ubuntu/Debian

curl -LsSf https://astral.sh/uv/install.sh | sh

Install the Graphify CLI

Graphify distributes via PyPI under the package name graphifyy (note the double-y), though the CLI command remains graphify. According to the pyproject.toml in the Graphify-Labs/graphify repository, this naming distinction prevents conflicts with other packages while maintaining intuitive CLI usage.

Install using uv (recommended for isolated environments):

uv tool install graphifyy

Alternative installation methods:

# Using pipx

pipx install graphifyy

# Using pip (requires manual PATH configuration)

pip install graphifyy

Fix PATH Issues

If your terminal cannot find the graphify command after installation, update your shell environment:

# For uv installations

uv tool update-shell

# For pipx installations

pipx ensurepath

Then restart your terminal session.

Register the Skill with Your AI Assistant

Once the CLI is available, register Graphify as a skill for your AI coding assistant. This writes the skill definition to platform-specific directories (e.g., .claude/skills/graphify/SKILL.md or .agents/skills/graphify/SKILL.md).

Global installation (default: Claude Code):

graphify install

Project-scoped installation (writes under current repository):

graphify install --project

The AGENTS.md file in your repository root defines platform-specific registration formats for alternatives like Codex or OpenCode.

Build Your First Knowledge Graph

With the skill registered, generate a knowledge graph for any codebase. Navigate to your project directory and run:

graphify .

This parses the repository locally using Tree-sitter ASTs—no data leaves your machine unless you enable optional semantic extraction features.

Graphify creates a graphify-out/ directory containing three artifacts:

• graphify-out/graph.html – Interactive visualization of the codebase structure
• graphify-out/GRAPH_REPORT.md – High-level summary of dependencies and modules
• graphify-out/graph.json – Machine-readable graph data for programmatic queries

Optional Configuration

Enable Auto-Rebuild on Commits

Install a Git hook to automatically regenerate the graph after every commit:

graphify hook install

This also registers a custom merge driver for graph.json that uses union-merge to resolve conflicts automatically.

Install Extras for Advanced Features

If you need PDF, Office document, video, or Neo4j support, install Graphify with specific extras. For example, to enable PDF text extraction:

uv tool install "graphifyy[pdf]"

Available extras are defined in the repository's pyproject.toml under [project.optional-dependencies].

Verify the Installation

Confirm your local Graphify setup works by querying a specific node from your generated graph:

graphify query "APIRouter"

You should see a structured report detailing the APIRouter node's connections and metadata, confirming that the graph.json index is accessible and properly formatted.

Summary

Setting up Graphify locally requires these key steps:

• Install Python 3.10+ and uv (or pipx) as your environment manager
• Install the graphifyy package using uv tool install graphifyy
• Fix PATH issues with uv tool update-shell if the command is not found
• Register the skill globally or per-project using graphify install
• Generate graphs locally with graphify . which populates graphify-out/ with HTML, Markdown, and JSON artifacts
• Optional: Configure Git hooks with graphify hook install for automatic rebuilds

Frequently Asked Questions

Why is the PyPI package named "graphifyy" instead of "graphify"?

The package name includes a double-y to avoid conflicts with existing packages on PyPI, while the CLI command installed to your PATH remains graphify for intuitive usage. This distinction is documented in the repository's README.md under the installation section.

Can I install Graphify using standard pip?

Yes, pip install graphifyy works, but you must manually ensure the graphify binary is in your system PATH. The uv and pipx methods are recommended because they handle environment isolation and PATH configuration automatically.

Does Graphify require an internet connection to work?

No. The core functionality runs entirely offline using local Tree-sitter AST parsing. Only optional features—such as semantic extraction from PDFs or media files—may call external LLM backends if you enable those specific extras.

What should I do if "graphify" command is not found after installation?

Run the appropriate PATH update command for your installation method: uv tool update-shell for uv installations or pipx ensurepath for pipx installations. Then restart your terminal. If issues persist, check that your Python scripts/bin directory is included in your system PATH environment variable.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}