LLM Discord Selfbot

A Discord selfbot that interfaces with any OpenAI-compatible LLM API to automatically reply to messages.

Note: This is a selfbot (runs on a user account, not a bot account). Use at your own risk; selfbots violate Discord's Terms of Service.

Features

• Mention replies: responds when @mentioned, with optional channel and user whitelists
• Random replies: probabilistically replies to 1 in every X messages
• Choice replies: periodically reviews recent messages and picks one to reply to
• DM support: always replies in 1-on-1 DMs; mention-only in group DMs (both independently toggleable)
• Vision support: understands images when using a vision-capable model
• Conversation memory: per-user conversation tracking that persists through channel noise
• Provider-agnostic: works with any OpenAI-compatible API (Anthropic, xAI/Grok, OpenAI, OpenRouter, etc.)
• Human-like typing delay: configurable simulated typing indicator
• Runtime commands: adjust settings on the fly via Discord messages

Setup

pip install -r requirements.txt

Copy config.example.yaml to config.yaml and fill in your tokens.

Usage

python bot.py

Commands

All commands use the configured prefix (default >). Only the selfbot's own user can trigger them.

Command	Description
>toggle mention	Enable/disable mention replies
>toggle random	Enable/disable random replies
>toggle interesting	Enable/disable interesting picker
>toggle dms	Enable/disable DM replies
>toggle groupdms	Enable/disable group DM replies
>status	Show current configuration
>model <name>	Switch LLM model
>prompt <text>	Update system prompt
>temp <X>	Set LLM temperature
>odds <N>	Set random reply odds (1 in N)

Configuration

See config.example.yaml for all options including:

• Channel and user ID whitelists for mention/random modes
• Separate context limits for DMs vs channels
• Vision toggle, typing delay, bot ignoring
• Interesting picker interval and lookback settings