Imported from facundososab/rapidfood (AGENTS.md). Install upstream with npx skills add facundososab/rapidfood. Copyright stays with the author.

Rapidfood — Agent Guidelines

Project Overview

Rapidfood is a backend application built in Python. It exposes an HTTP shell using Django 5 and Django REST Framework (DRF), while the data layer is completely delegated to Prisma Client Python. The business logic is distributed across independent backend modules following Hexagonal Architecture.

Repository Guidelines

How to Use This Guide

Start here for cross-project norms. This is the main configuration file for AI agents working on Rapidfood.

Component docs override this file when guidance conflicts. Always refer to docs/ARCHITECTURE-GUIDE.md for in-depth architectural decisions.

Key Cross-Project Rules

• Language: All code, comments, variable names, and commit messages should be in English; business domain terms and UI copy may be in Spanish to match the ubiquitous language.
• Package manager: uv at the root. Dependencies are in pyproject.toml. Do NOT use or generate requirements.txt.
• Database Rules: Prisma (schema.prisma) is the absolute source of truth for the database. Django ORM is only a bridge, never the owner.
• Hexagonal Architecture: Dependencies flow inwards (infrastructure -> application -> domain).
• Never commit secrets: Use environment variables; .env files are gitignored.
• Architecture Enforcement: Run import-linter to verify cross-app boundaries and layer constraints before merging.

Agent Skills

Auto-invoke Skills

When performing these actions, ALWAYS invoke the corresponding skill FIRST: