setup-graphify

A Claude Code skill that installs and optimizes graphify knowledge graph for any project with a single command.

What it does

Type /setup-graphify in any project and it will:

1. Install graphifyy Python package (if needed)
2. Configure Claude Code integration (CLAUDE.md + PreToolUse hook)
3. Initialize git and install post-commit/post-checkout hooks
4. Build the full knowledge graph (AST extraction + clustering + report)
5. Generate interactive HTML visualization
6. Set up daily scheduled auto-rebuild

Install

Copy the skill to your Claude Code skills directory:

# Windows
mkdir %USERPROFILE%\.claude\skills\setup-graphify
copy SKILL.md %USERPROFILE%\.claude\skills\setup-graphify\SKILL.md

# macOS / Linux
mkdir -p ~/.claude/skills/setup-graphify
cp SKILL.md ~/.claude/skills/setup-graphify/SKILL.md

Or use the install script:

# Windows (PowerShell)
.\install.ps1

# macOS / Linux
chmod +x install.sh && ./install.sh

Usage

Open Claude Code in any project directory and type:

/setup-graphify

That's it. Everything is automatic.

Prerequisites

• Python 3.10+
• Claude Code

What you get after setup

your-project/
├── .claude/
│   └── settings.json          # PreToolUse hook
├── .git/
│   └── hooks/
│       ├── post-commit        # auto-rebuild on commit
│       └── post-checkout      # auto-rebuild on branch switch
├── .graphifyignore            # excluded paths
├── CLAUDE.md                  # graphify rules + architecture summary
└── graphify-out/
    ├── GRAPH_REPORT.md        # god nodes, communities, knowledge gaps
    ├── graph.json             # queryable graph data
    └── graph.html             # interactive visualization

How it works with Claude Code

After setup, Claude Code will:

• Automatically consult the knowledge graph before searching raw files (via PreToolUse hook)
• Automatically rebuild the graph after each git commit (via git hook)
• Automatically rebuild daily via scheduled task

You don't need to change how you work. Just give tasks as usual.

Credits

• graphify by safishamsi
• Inspired by Andrej Karpathy's knowledge base workflow