How gittodiagram/gittodiagram-be works
Project type and purpose
• Frontend = Next.js + TypeScript + Tailwind;
Backend = FastAPI + Python;
Database = PostgreSQL;
AI services = OpenAI / Claude;
Proxy = Nginx;
Orchestration = Docker Compose;
Deploy = Vercel (frontend), EC2 (backend).
File structure
Top-level:
• Dockerfile / docker-compose.yml / entrypoint.sh → container definitions for backend, DB, Nginx
• nginx/ → reverse-proxy conf (api.conf)
• start-database.sh → launch Postgres
• app/ → FastAPI application
– app/main.py → entry point, mounts routers & middleware
– app/routers/*.py → API endpoints (generate, modify)
– app/services/*.py → wrappers for AI providers & GitHub
– app/core/limiter.py → rate-limiting logic
– app/prompts.py → prompt-engineering pipeline
– app/utils/format_message.py → post-processing responses
Extract README insights
• Frontend and backend are separated (hosted on different platforms) but communicate via HTTP.
• PostgreSQL (via Drizzle ORM) persists any usage data, if applicable.
• AI calls are abstracted behind service classes.
• Docker + Nginx sit in front of the FastAPI server in production.
Don’t overcomplicate—capture every distinct service or layer.
– Optionally show CI/CD → deployment targets (Vercel, EC2).