Maquette

Overview

Maquette is an AI-powered deployment platform that takes any GitHub repository and deploys it to production infrastructure automatically. Users provide a repository URL and optional environment variables, and an AI agent takes over — analyzing the codebase, generating Dockerfiles and docker-compose configurations, configuring Traefik reverse proxy routes, setting up Cloudflare DNS records, and deploying the service. The entire process streams in real time through a structured deployment log with expandable phases. Once deployed, each service gets its own dashboard with uptime monitoring, response time charts, web analytics, custom domain management, and one-click updates. A Discord bot integration provides deployment notifications and remote management capabilities.

Tech Stack

Technical Details

The platform consists of four services: a FastAPI backend, a React Router 7 frontend, an uptime monitoring service, and a web analytics processor.

The backend orchestrates deployments through a custom AI agent framework built on Google's Gemini API with tool-use capabilities. The agent executes bash commands in the cloned repository directory, iterating through planning, Dockerfile generation, docker-compose setup, Traefik route configuration, and Cloudflare DNS provisioning. Status updates stream to the frontend via Server-Sent Events (SSE), with each update stored in MongoDB for persistence. The agent uses a structured status reporting system with nested phases, tool calls, and error tracking.

The frontend is a React 19 SPA with React Router 7 and Vite, using Auth0 for authentication. The deployment view renders a real-time log of AI agent actions organized into collapsible phases. The service hub displays deployed services as cards with live status indicators. Each service detail view includes uptime monitoring with response time charts (Recharts), web analytics panels, custom domain management with automatic Cloudflare DNS and Traefik configuration, and deployment log history.

The uptime monitoring service runs periodic health checks against all deployed services, stores response time data in MongoDB, and captures screenshots via a headless browser. The web analytics processor parses Traefik access logs to provide per-service traffic metrics without client-side tracking scripts.