ArchitectOS

A local-first AI architecture visualizer. Describe any system and watch it decompose into an interactive, drillable graph — or upload your existing codebase to visualize it.

Powered by Ollama (free, local AI). No API keys. No billing. No cloud.

Frontend: Vue 3 + Vite Backend: Express.js (TS) with integrated Ollama AI

✨ Features

Core

• AI Decompose — describe any system, AI generates a multi-level architecture graph
• Hybrid Mode — AI generates the base, then edit/add/delete nodes inline
• Drill-down — click any node to zoom into its children; breadcrumb navigation back
• Syntax Toggle — hide code, show pseudocode, or show real implementation code

Autonomy Levels

Level	Depth	What you get
Beginner	1	High-level system overview
Intermediate	2	Subsystems + data flow
Advanced	3	Internal mechanisms, auth, protocols
Expert	4	Full implementation detail

Tools

• 📝 README Generator — auto-generates a professional README from your architecture; turns red when stale
• 📁 File Structure — view the generated project's file tree
• 📦 Download ZIP — download everything (code, README, config) as a ready-to-use project
• 📤 Upload Codebase — upload a ZIP or source files to reverse-engineer into a flowchart
• 🗺️ Minimap — birds-eye view of the full graph
• 🔎 Zoom Controls — zoom in/out and fit-to-view

Quality of Life

• Persistent state — refresh the page and everything is still there (localStorage)
• Prompt history — quickly re-run recent prompts
• Error handling — clear error messages with dismiss button
• No overlap — tree-aware layout allocates space proportionally

Requirements

• Node.js 18+
• Ollama with a model installed

Quick Start

# 1. Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# 2. Pull a model
ollama pull llama3.1:8b

# 3. Clone & install
git clone https://github.com/ThatDeparted2061/ArchitectOS.git
cd ArchitectOS
npm install

# 4. Run
npm run dev

Open http://localhost:5173

First generation takes 15-30s (model loading into RAM). Subsequent ones are faster.

Configuration

Edit backend/.env:

AI_ENABLED=true
OLLAMA_BASE_URL=http://localhost:11434
OLLAMA_MODEL=llama3.1:8b

Swap OLLAMA_MODEL for any model: mistral, codellama, llama3.1:70b, etc.

Modes

Mode	Description
AI Decompose	Prompt → AI generates full architecture
Hybrid	AI generates → you edit, add, or delete nodes
Manual Mode	Build from scratch (coming soon)

How Upload Works

1. Click 📤 Upload Codebase
2. Select source files or a ZIP archive
3. AI analyzes the code structure and generates an architecture flowchart
4. Edit in Hybrid mode, download as ZIP, or generate a README

Supports: .js, .ts, .py, .java, .go, .rs, .c, .vue, .svelte, and 30+ other file types.

Stack

Layer	Technology
Frontend	Vue 3, Vite, Vue Flow, Pinia, Tailwind CSS
Backend	Express, TypeScript
AI	Ollama (local LLM)

Project Structure

architect-os/
├── backend/
│   ├── index.ts          # Express API server
│   ├── mock/             # Fallback mock data
│   └── .env              # Configuration
├── frontend/
│   ├── src/
│   │   ├── components/   # Vue components
│   │   ├── services/     # API + graph layout
│   │   ├── store/        # Pinia state management
│   │   └── App.vue       # Main app
│   └── ...
└── package.json          # Workspace root

License

MIT