Liquid Mod

A Discord bot made to prevent scams on the Liquid AI Community Server

---

Features

• Keyword moderation — automatically delete, kick, ban, or mute users based on configurable word lists
• AI scam detection — uses a local LLM to flag and delete likely scam messages above a character threshold
• Debug mode — when enabled, the bot processes all messages including those from admins and itself

---

Prerequisites

• Python 3.9+
• llama-cpp-python installed with GPU support (recommended) or CPU
• A Discord bot token with the Message Content Intent enabled

---

Installation

Windows

REM 1. Clone the repository
git clone <your-repo-url>
cd <repo-folder>

REM 2. Create and activate a virtual environment
python -m venv venv
venv\Scripts\activate

REM 3. Install llama-cpp-python (CPU only - for GPU support install CUDA and use cu124 wheel)
pip install llama-cpp-python

REM 4. Install Python dependencies
pip install discord.py python-dotenv tomli

REM 5. Download the model
REM Download LFM2.5-1.2B-Instruct-Q4_K_M.gguf from:
REM https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF
REM Place it in a "models" folder in the project root

Ubuntu (with llama-cpp-python)

# 1. Clone the repository
git clone <your-repo-url>
cd <repo-folder>

# 2. Install system dependencies
sudo apt update
sudo apt install -y build-essential cmake git libcurl4-openssl-dev python3-dev python3-venv

# 3. Create and activate a virtual environment
python3 -m venv venv
source venv/bin/activate

# 4. Install llama-cpp-python
# For CUDA GPU support:
pip install llama-cpp-python --force-reinstall --no-cache-dir \
    --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cu124

# For CPU only (slower, no GPU):
# pip install llama-cpp-python

# 5. Install Python dependencies
pip install discord.py python-dotenv tomli

# 6. Create models directory and download the model
mkdir -p models
cd models

Unix (Linux / macOS)

# 1. Clone the repository
git clone <your-repo-url>
cd <repo-folder>

# 2. Install system dependencies (Linux only)
# For Ubuntu/Debian:
sudo apt update
sudo apt install -y build-essential cmake git libcurl4-openssl-dev python3-dev python3-venv

# 3. Create and activate a virtual environment
python3 -m venv venv
source venv/bin/activate

# 4. Install llama-cpp-python (CPU only - for GPU see Ubuntu section)
pip install llama-cpp-python

# 5. Install Python dependencies
pip install discord.py python-dotenv tomli

Setting up the AI

The bot requires the LFM2.5-1.2B-Instruct-Q4_K_M.gguf model file. Download it from HuggingFace:

# Create models directory
mkdir -p models

# Download using huggingface-cli
pip install huggingface-hub
huggingface-cli download LiquidAI/LFM2.5-1.2B-Instruct-GGUF LFM2.5-1.2B-Instruct-Q4_K_M.gguf --local-dir models

Or manually from: https://huggingface.co/LiquidAI/LFM2.5-1.2B-Instruct-GGUF

However, if you want to use your own model, you can customize it inside the config.toml file.

Creating the bot

---

Configuration

1. Create a .env file in the project root:

   BOT_TOKEN=your_discord_bot_token_here
   

2. Edit config.toml to configure moderation rules and AI settings:

   • Add words/phrases to the delete, kick, ban, or mute lists under [moderation]
   • Adjust time_to_mute (in minutes) for timed-out users
   • Set min_chars under [ai] to control the minimum message length the AI will check
   • Set debug_mode = false under [debug] when deploying to production

---

Running the Bot

Make sure your GGUF model file is in place (see model_path in config.toml), then start the bot:

python main.py