Zappy - Epitech Project

Goal

The goal of the Zappy project is to build an ecosystem composed of 3 distinct binaries:

• Server: Runs the entire game world and rules.
• GUI: Visualizes the map, players, and events in real-time.
• AI: Controls autonomous drones competing to survive and evolve.

Launch

To clone, build, and launch the Zappy project, please follow the instructions below:

1. Clone the repository Do not forget to use --recurse-submodules as this project relies heavily on several internal submodules:

   git clone [https://github.com/EpitechPGE2-2025/G-YEP-400-NCY-4-1-zappy-3](https://github.com/EpitechPGE2-2025/G-YEP-400-NCY-4-1-zappy-3) zappy/ --recurse-submodules -j8
   

2. Build the project with CMake

   cmake -B build
   cmake --build build
   

   2.1 Alternatively, you can use Bazel (available on the feature/bazel-support branch):

   bazel build //server:zappy_server
   bazel build //gui:zappy_gui
   cd bazel-bin/
   

3. Launch the different components

   3.1 Launch the Server:

   ./zappy_server -p 4242 -n team1 team2 -f 100 -c 10 -v 3
   

   3.2 Launch the GUI:

   ./zappy_gui -m localhost -p 4242
   

   3.3 Launch the AI:

   cd IA/
   python3 zappy_ai.py -h localhost -p 4242 -n team1
   

---

Server

The zappy_server is a tick-based application designed to ensure absolute deterministic simulation. Because actions are calculated using server-side ticks rather than delta-time, physics glitches or movement speed exploits caused by client lag are completely eliminated.

The server hosts a high-performance network layer allowing asynchronous TCP communication between all connected clients (AIs and GUIs):

• The GUI continuously polls and retrieves state commands to display the world.
• The AIs send action commands (e.g., Forward, Look, Take) to interact with the environment.

Core Libraries Used

• Kronknet: A custom, optimized network library used to manage TCP connections. It provides clean callback mechanisms (on connect, on read, etc.) and exposes a userPtr context payload, which enables effortless retrieval of player contexts directly inside event hooks.
• Prophecy: A lightweight task-scheduler library. When an AI triggers a delayed action like Fork, the server creates a delayed task and hands it over to Prophecy, scheduling its execution exactly 42 ticks into the future.
• Argot: Used for flexible command-line argument parsing. It sets up strict options (like --port (-p)) and safely extracts configuration fallback constraints (e.g., get("port", default_value=4242)).

Submodules:

• kronknet — Custom Event-Driven Socket Engine.
• prophecy — Tick-Based Priority Task Scheduler.
• argot — Strict Argument Parsing Utility.

---

GUI

The zappy_gui application displays the game universe graphically.

Key Features:

• Custom Entity-Component-System (ECS) architecture for efficient rendering updates.
• Pure OpenGL graphics context pipeline.
• Custom software implementation of Rasterization mechanics.

---

AI

The zappy_ai application acts as an autonomous client designed to parse instructions, manage its inventory, and make survival decisions using state-machine algorithms.

---

Contributing

• Server Module:
  • Clément Devaux
  • Gauthier Malfilatre
• GUI Module:
  • Julian Hemmer
  • Romain Marcot
• AI Module:
  • Martin Muller
  • Victor Zhou

© 2026 NIFH - National Institute for Féeria History