Reference

Reference Guide

Reference pages provide a concise overview of how the Discord AI Bot works and the main settings available in the project.

Core Components

Component	Purpose
Discord_bot.py	Main Python script
Ollama	Runs the local AI model
Playwright	Controls the browser interface
Chromium	Browser used by Playwright
requirements.txt	Python dependencies

AI Model

The bot uses Ollama to generate responses.

The default model configured in the script is:

model='qwen2.5:14b'

The model can be changed to another Ollama model, provided that model has been installed locally.

Example:

ollama pull qwen2.5:7b

Then update the model name in Discord_bot.py.

Blacklist

The BLACKLIST variable contains usernames or words that should be filtered.

BLACKLIST = ["yo"]

Multiple entries can be specified:

BLACKLIST = ["example", "ExampleUser", "AnotherUser"]

The blacklist is checked when processing usernames and generated responses.

Mention Blacklist

The MENTION_BLACKLIST variable controls which users should not be automatically mentioned.

MENTION_BLACKLIST = ["yes", "bro"]

Users on this list can still receive responses, but the automatic mention is skipped.

AI Personality

The normal AI behavior is controlled by the system_prompt.

The default personality is designed to be:

• Knowledgeable about technology
• Familiar with Windows, Linux, macOS, hardware, and scripting
• Short and conversational
• Helpful and technically focused
• Occasionally sarcastic

The system prompt can be edited directly in Discord_bot.py to customize the bot’s personality.

Fourth-Wall Roast Mode

The bot includes a random fourth-wall roast mode.

The default probability is:

is_roast = random.random() < 0.25

This gives approximately a 25% chance of triggering roast mode.

Common values:

Value	Approximate probability
0.10	10%
0.25	25%
0.50	50%
0.75	75%

The roast target is configured separately:

target_user_to_tag = "exampleuser"

Link Detection

The bot checks messages for URLs using:

url_match = re.search(r'(https?://[^\s]+)', latest_message)

When a URL is detected, the script attempts to open the page and extract its visible text.

The extracted page content is limited by:

page_text_clean = " ".join(raw_text.split())[:1500]

The 1500 value controls the maximum number of characters included in the AI prompt.

Response Processing

Generated responses are cleaned before being sent.

Quotation marks are removed with:

reply_text = reply_text.replace('"', '').replace("'", '')

The generated response is also checked against the configured blacklist.

Blacklisted words are replaced with:

the user

Dependencies

The project uses Python packages including:

ollama
playwright

Python’s built-in modules asyncio, random, and re are also used.

Install dependencies with:

pip install -r requirements.txt

Install Playwright’s Chromium browser with:

playwright install chromium

Runtime

The main script is started with:

python Discord_bot.py

The selected Ollama model must be installed before starting the application.

Security

Never commit passwords, API keys, authentication tokens, session cookies, or other private credentials to GitHub.

Sensitive configuration should be stored using environment variables or another secure method.