🎮 The Ultimate AI-Powered Bot for MMORPG and RPG Games 🤖
Farm resources, level up, and dominate your favorite games while you sleep!
Features • Quick Start • See It In Action • Setup Guide • Support
Tired of grinding for hours? Let AI do it for you!
BOT-MMORPG-AI is your personal gaming assistant that uses artificial intelligence to play your favorite MMORPG and RPG games automatically. It watches how YOU play, learns from your gameplay, and then takes over the boring, repetitive tasks while you relax, work, or sleep.
- Genshin Impact (Primary Support)
- New World
- World of Warcraft
- Guild Wars 2
- Final Fantasy XIV
- Elder Scrolls Online
- And many more!
Unlike simple macro bots that just repeat actions, this AI actually learns how to play by watching you! It uses the same technology behind self-driving cars and facial recognition to understand your game and make smart decisions in real-time.
- ✅ Auto-Farming: Collect resources automatically while you're away
- ✅ Smart Navigation: Travel from point A to point B without getting stuck
- ✅ Combat AI: Fight enemies and complete dungeons autonomously
- ✅ Item Collection: Never miss loot again
- ✅ 24/7 Operation: Farm even when you're sleeping
- ✅ Human-Like Behavior: Plays naturally, not like a robot
- ✅ Controller Support: Works with keyboard AND gamepad
- 🎯 Learns From You: Records YOUR gameplay and mimics YOUR style
- 🎯 Adaptive AI: Gets smarter the more you train it
- 🎯 Stuck Detection: Automatically escapes when trapped
- 🎯 Path Recognition: Knows where to go using computer vision
- 🎯 Multi-Action Support: Handles complex button combinations
Navigate from Mondstadt to Thousand Wind Temple automatically:
The bot learns the path and can travel autonomously
Set the perfect in-game time for your farming routes
Watch the AI play in real-time with natural, human-like movements
Full support for controllers and gamepads
Advanced computer vision identifies safe paths and obstacles
High accuracy training results - the AI learns quickly!
Don't worry - you don't need to be a programmer! We've made this super simple.
# Just copy and paste these commands one by one
curl -LsSf https://astral.sh/uv/install.sh | sh
git clone https://github.com/ruslanmv/BOT-MMORPG-AI.git
cd BOT-MMORPG-AI
make installThat's it! The bot is now installed.
- Open your game (Genshin Impact recommended)
- Set your resolution to 1920x1080 fullscreen
- Run:
make collect-data - Play normally for 10-15 minutes
- The bot is now learning!
make train-modelGrab a coffee ☕ - training takes 30-60 minutes depending on your GPU.
make test-modelBoom! Your AI is now playing for you! 🎉
📖 Need more help? Check out our detailed setup guide for gamers!
- PC: Windows 10/11 (where you play your game)
- Python: 3.8 or newer (we'll help you install it)
- Space: 5GB free disk space
- RAM: 8GB minimum, 16GB recommended
- Game: Any supported MMORPG/RPG game
- GPU: NVIDIA Graphics Card (makes training 10x faster!)
- Controller: Xbox/PS controller (optional, but recommended)
- Internet: For downloading dependencies
- ✅ Genshin Impact (Best Support)
- ✅ New World
- ✅ World of Warcraft
- ✅ Guild Wars 2
- ✅ Final Fantasy XIV
- ✅ Elder Scrolls Online
- ✅ Most other MMORPG/RPG games!
See our Gamer's Setup Guide - step-by-step with screenshots!
Click to expand installation commands
# Quick Install with UV (Recommended)
curl -LsSf https://astral.sh/uv/install.sh | sh
git clone https://github.com/ruslanmv/BOT-MMORPG-AI.git
cd BOT-MMORPG-AI
make install-all
# Traditional Install
git clone https://github.com/ruslanmv/BOT-MMORPG-AI.git
cd BOT-MMORPG-AI
python -m venv venv
venv\Scripts\activate # On Windows
pip install -e ".[dev]"
Stand at the bridge of Mondstadt - this is your starting point!
Set in-game time to 12:00 for consistent lighting and better AI performance
If using a gamepad, make sure it's connected and recognized
make collect-data- Play normally for 10-15 minutes
- The AI watches and learns from you
- Press
Tto pause/resume recording - Press
Qto stop and save
💡 Pro Tip: Play the same route 2-3 times for better results!
make train-modelThis trains the neural network on your gameplay. Time depends on your hardware:
- With GPU: 30-60 minutes
- Without GPU: 2-4 hours
☕ Perfect time for a coffee break!
make test-model- Position your character at the bridge
- Set time to 12:00
- Run the command
- Switch to game window
- Watch the magic happen! ✨
Controls while bot is playing:
T: Pause/resume the AIESC: Stop the bot completely
The BOT-MMORPG-AI Launcher is a gaming-style dashboard that lets you manage the entire AI lifecycle without touching the command line. Built with Eel, it acts as a central control panel where you can collect data, train your models, and run the bot with a single click.
It features real-time terminal feedback, automatic API key loading, and process management to start or stop tasks instantly.
First, install the launcher dependencies (specifically Eel and Chromium wrappers):
make install-launcher
Ensure you have your API keys set up. Create a .env file in the project root (see .env.example) and add your Google Gemini API key:
GEMINI_API_KEY=your_key_here
Launch the dashboard using the following command:
uv run python launcher/launcher.py
- 🔴 Start Recording: Runs
1-collect_data.pyto capture screen data and user inputs for the dataset. - 🧠 Train Model: Runs
2-train_model.pyto process the data and train the deep learning model. You will see real-time training logs in the launcher terminal. ▶️ Start Bot: Runs3-test_model.pyto activate the AI agent and let it play the game autonomously.
Coming soon! Subscribe to our YouTube channel for the full walkthrough.
For now, check our detailed written guide with screenshots!
BOT-MMORPG-AI/
├── src/
│ └── bot_mmorpg/ # Main package source code
│ ├── __init__.py
│ ├── models/ # Neural network architectures
│ ├── utils/ # Utility functions
│ └── scripts/ # Entry point scripts
├── versions/
│ └── 0.01/ # Version-specific implementations
│ ├── 1-collect_data.py
│ ├── 2-train_model.py
│ ├── 3-test_model.py
│ ├── models.py # Model definitions
│ ├── grabscreen.py # Screen capture
│ ├── getkeys.py # Keyboard input
│ ├── getgamepad.py # Gamepad input
│ └── directkeys.py # Key simulation
├── frontend/
│ ├── input_record/ # Input recording utilities
│ └── video_record/ # Video recording utilities
├── tests/ # Test suite
├── assets/ # Images and resources
├── pyproject.toml # Project configuration
├── Makefile # Build automation
├── LICENSE # Apache 2.0 License
└── README.md # This file
# Format code
make format
# Run linters
make lint
# Type checking
make type-check
# Run all checks
make check# Run all tests
make test
# Run with coverage
make test-cov
# Run specific test types
make test-unit # Unit tests only
make test-integration # Integration tests only
make test-fast # Exclude slow tests# Build distribution packages
make build
# Generate documentation
make docs
# Full CI pipeline
make ciThe project supports multiple neural network architectures:
| Model | Size | Top-1 Acc | Top-5 Acc | Parameters | Inference (GPU) |
|---|---|---|---|---|---|
| InceptionV3 | 92 MB | 77.9% | 93.7% | 23.9M | 6.9 ms |
| ResNet50 | 98 MB | 74.9% | 92.1% | 25.6M | 4.6 ms |
| ResNet101 | 171 MB | 76.4% | 92.8% | 44.7M | 5.2 ms |
| VGG16 | 528 MB | 71.3% | 90.1% | 138.4M | 4.2 ms |
| MobileNetV2 | 14 MB | 71.3% | 90.1% | 3.5M | 3.8 ms |
The model predicts 29 different actions:
- Keyboard (9): W, S, A, D, WA, WD, SA, SD, NOKEY
- Gamepad (20): LT, RT, Lx, Ly, Rx, Ry, D-Pad, Buttons (A, B, X, Y, etc.)
Data is stored as NumPy arrays:
- Input: Screen captures (480x270x3 RGB images)
- Output: Multi-hot encoded action vectors (29 classes)
- Format:
.npyfiles with 500 samples each
For interactive development and experimentation:
# Install Jupyter dependencies
pip install -e ".[jupyter]"
# Launch JupyterLab
jupyter labAvailable notebooks in versions/0.01/:
- Data collection and preprocessing
- Model training with visualizations
- Data cleaning and augmentation
- Way identification using OpenCV
- Intermediate representation visualization
The project supports cloud-based training on:
- Google Colab: Free GPU training
- AWS EMR: Scalable cluster training
- Azure ML: Enterprise-grade training
- Google Cloud AI Platform: Distributed training
- U-Net Models: For semantic segmentation of game paths
- LSTM Networks: For temporal action prediction
- ResNeXt: Advanced residual network architectures
- 3D Convolutions: For multi-frame temporal learning
The bot uses motion detection to prevent getting stuck:
motion_req = 800 # Minimum motion threshold
log_len = 25 # Motion history lengthWhen stuck, the bot executes random evasive maneuvers.
Custom weights applied to predictions for game-specific behavior:
weights = [4.5, 0.1, 0.1, 0.1, 1.8, 1.8, 0.5, 0.5, 0.2, ...]Issue: Model not loading
- Solution: Ensure model files are in
model/directory
Issue: Screen capture not working
- Solution: Run game in fullscreen 1920x1080 resolution
Issue: Bot getting stuck
- Solution: Adjust
motion_reqthreshold in test script
Issue: Low FPS during recording
- Solution: Lower capture resolution or use faster storage
Issue: CUDA out of memory
- Solution: Reduce batch size in training configuration
We welcome contributions from the community!
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
- Follow PEP 8 style guidelines
- Add docstrings to all functions and classes
- Include type hints for function signatures
- Write unit tests for new features
- Update documentation as needed
- Run
make checkbefore submitting
Please read our Code of Conduct before contributing.
Get Help, Share Tips, Show Off Your Bots!
- 💬 Slack: #bot-mmorpg-ai - Live chat with other gamers
- 🐛 Report Issues: GitHub Issues - Found a bug?
- 🌐 Website: ruslanmv.com - Tutorials and guides
- 📧 Email: contact@ruslanmv.com - Direct support
- ⭐ Star this repo to get updates
- 👁️ Watch for new releases
- 🔔 Subscribe to our YouTube (coming soon!)
This project builds upon excellent work from the community:
- gamePyd - Game control utilities
- vJoy - Virtual joystick interface
- ScpVBus by nefarius
- PYXInput contributors
- PyGTA5 by Sentdex
- Inception V3 architecture by Google Research
Special thanks to the AI Gaming Community for their continuous support and feedback.
- Multi-game support framework
- Web-based dashboard for monitoring
- Improved data augmentation pipeline
- Distributed training support
- Reinforcement learning integration
- Real-time model updating
- Cloud storage integration (S3, MinIO)
- Performance profiling tools
- Generalized game agent framework
- Plugin system for custom games
- Advanced reward shaping
- Model compression for edge deployment
If you use this project in your research, please cite:
@software{magana2025botmmorpgai,
author = {Magana Vsevolodovna, Ruslan},
title = {BOT-MMORPG-AI: AI-Powered Bot for MMORPG Games},
year = {2025},
url = {https://github.com/ruslanmv/BOT-MMORPG-AI},
version = {1.0.0}
}This project is licensed under the Apache License 2.0 - see the LICENSE file for details.
Copyright 2025 Ruslan Magana Vsevolodovna
Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at
http://www.apache.org/licenses/LICENSE-2.0
Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.
Ruslan Magana Vsevolodovna
- Website: ruslanmv.com
- Email: contact@ruslanmv.com
- GitHub: @ruslanmv
Made with ❤️ by the AI Gaming Community