Archon

The greated AI cloud architect, so you can kickback and relax.

Features

• Simplifies process of designing cloud architectures
• Picks architectural patterns
• Finds the best services from AWS, Azure and GCP using RAG
• Validates the architecture against the well-architected framework
• Generates Diagrams via an MCP server
• Fully responsive UI
• User authentication baked in via Better-Auth

Demo video

You can see a demo video here

Architecture

• We first gathered data from all 3 cloud providers, and assigned tiers to them (T1/T2/T3)
• During Phase 2, we perform RAG only on T2 and T3 docs. T1 docs are always injected into the context, since they are foundational/common services
• The pipeline does the following in order:
  • Phase 0: Gathers app requirements
  • Phase 1: Finds best architectural patterns
  • Phase 2: Finds best services from AWS, Azure and GCP in parallel (split into 2 waves, since networking, devops and security depend on wave 1's output)
  • Phase 3: Validates the architecture using the well-architected framework
  • Phase 4: Creates a final response, taking into account all of the specialist agent's output
  • Phase 5: Generates Diagram via an MCP server
• Overview of the main pipeline:

• When the user wants to ask a follow-up question, it follows a different pipeline:

Prerequisites

App

• Node.js
• pnpm
• PostgreSQL database

Diagram generation

• Python 3.12+
• uv
• Graphviz:

sudo apt-get install graphviz graphviz-dev

Running locally

1. Install Next.js dependencies

pnpm install

2. Configure environment variables

cp .env.example .env

Fill in the following values:

• OPENAI_API_KEY — API key for OpenAI
• DATABASE_URL — PostgreSQL connection string, e.g. postgres://user:password@localhost:5432/archon
• BETTER_AUTH_SECRET — Random secret used to sign auth tokens. You can use this website, or run:

openssl rand -base64 32

3. Run database migrations

pnpm db:migrate

4. Start the MCP diagram server

The diagram generation feature requires the MCP server running separately. In a new terminal:

cd mcp
uv sync
uv run diagram-mcp

The server starts on http://localhost:8000 by default. The Next.js app connects to it via the MCP_SERVER_URL environment variable.

5. Start the Next.js development server

pnpm dev

Open http://localhost:3000 in your browser.

Via Docker

You can also run Archon directly via docker.

Make sure the .env file is configured:

cp .env.example .env

• OPENAI_API_KEY — API key for OpenAI
• DATABASE_URL — Ignore this, will be configured automatically
• BETTER_AUTH_SECRET — Random secret used to sign auth tokens. You can use this website, or run:

openssl rand -base64 32

then run:

docker compose build
docker compose up

Open http://localhost:3000 in your browser.