DEV Community

Jervi
Jervi Subscriber

Posted on

Graphify on macOS, to shorten the LLM consumption

#graphify #llm #markdown

Article in my blog

Quick Start overall

| install the following

brew install python@3.12 uv

uv tool install graphifyy
Enter fullscreen mode Exit fullscreen mode

then reload the shell

then run the following

uv tool update-shell
source ~/.zshrc
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc
Enter fullscreen mode Exit fullscreen mode

Run Graphify:

graphify
Enter fullscreen mode Exit fullscreen mode

Run the MCP server:

graphify-mcp
Enter fullscreen mode Exit fullscreen mode

done

Why Graphify?

Graphify helps create a structured knowledge graph of a codebase. Instead of forcing an LLM to rediscover architecture, relationships, modules, and patterns every time, Graphify builds a persistent representation of the project.

This is especially useful for large NestJS, Next.js, ERP, and monorepo projects where AI assistants tend to lose context or regenerate incomplete understanding after long conversations.

Benefits:

  • Faster AI onboarding to a project
  • Better architectural understanding
  • Reduced repeated explanations
  • More consistent code generation
  • Persistent project memory across sessions

How I Use It

My goal is not just documentation generation.

I use Graphify to create a persistent understanding of the codebase so AI agents can work from an existing project memory instead of rebuilding their understanding from scratch every session.

This is particularly useful for:

  • NestJS backends
  • Next.js frontends
  • ERP systems
  • Monorepos
  • Long-running projects with many modules

The result is better continuity, fewer hallucinated architectural decisions, and less time spent re-explaining the project to AI tools.

Top comments (0)