llmcord

llmcord is a self-hosted Discord bot for people who want to share AI conversations with friends or a community. It connects Discord chats to local models or remote providers. Conversation history stays in Discord, even when the model runs on your own hardware.

• Shared conversations: Branch a discussion, continue someone else's conversation or carry a branch into a thread, with Discord reply chains defining the context the model sees.
• Model choice: Connect to Ollama, LM Studio or vLLM for local inference, or use OpenRouter, OpenAI, xAI and Google through compatible APIs, then switch between configured models within Discord.
• File and image input: Include text and code attachments in a conversation, plus images when the selected model supports vision.
• Chat behavior: Set a custom system prompt and receive streamed answers, with long responses split across messages; the bot distinguishes participants by their Discord IDs and also supports direct messages.

The Python bot is open source under the MIT license. Backends must support the OpenAI-compatible chat completions API. Local backends process conversations on your hardware; remote backends send the conversation context to the selected provider. Discord stores the reply history, so the bot doesn't require a separate database. A configurable history limit drops the oldest messages when a conversation exceeds it.