jakobdylanc/llmcord

What it solves

llmcord turns a Discord server into a collaborative interface for interacting with Large Language Models (LLMs). It allows groups of friends or community members to chat with AI together, branching conversations and sharing context without needing a separate standalone frontend or a complex database setup.

How it works

The bot integrates with Discord and connects to any OpenAI-compatible API (including remote providers like OpenRouter and Google, or local setups like Ollama and LM Studio). It uses Discord's own reply system to track conversation history; by replying to a message, the bot treats that entire reply chain as the context for the current prompt. This eliminates the need for an external database to store chat history.

Who it’s for

Users who want to bring LLM capabilities into their Discord communities for collaborative brainstorming, shared AI assistance, or social experimentation with different models.

Highlights

• Reply-based Context: Uses Discord reply chains to manage conversation history and branching.
• Broad Model Support: Compatible with any OpenAI /v1/chat/completions API, including local and remote providers.
• Multimodal Capabilities: Supports image and text file attachments when using vision-capable models.
• Flexible Configuration: Features hot-reloading settings, customizable system prompts, and granular permission controls for users, roles, and channels.
• Efficient Design: Fully asynchronous implementation in a single Python file with optimized message caching.