discord-llms

A Discord bot that uses LLMs to help users with project documentation. Point it at llms.txt files and let users ask questions in Discord threads.

Features

• Thread-based conversations: Each question starts a new thread for focused discussion
• Smart document selection: LLM automatically selects relevant documentation for each question
• Multi-doc support: Configure multiple documentation sources with summaries
• OpenAI-compatible API support: Use any OpenAI-compatible provider (OpenRouter, OpenAI, Anthropic, local LLMs, etc.)
• Tool calling: Uses LLM tool calling to intelligently select documentation

How it works

1. User mentions the bot with a question: @docs-bot How do I use tool calling?
2. Bot creates a new thread for the conversation
3. LLM analyzes the question and selects relevant documentation using tool calling
4. Bot fetches the selected llms.txt files
5. LLM answers the question using the documentation context
6. Follow-up questions in the thread continue the conversation

Installation

Install uv:

Linux/Mac:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows (PowerShell):

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

Configuration

Configuration File

Download the example configuration file:

Linux/Mac:

curl -o config.yaml https://raw.githubusercontent.com/S1M0N38/discord-llms.txt/main/config/config.example.yaml

Windows (PowerShell):

Invoke-WebRequest -Uri https://raw.githubusercontent.com/S1M0N38/discord-llms.txt/main/config/config.example.yaml -OutFile config.yaml

Discord Bot Setup

1. Go to Discord Developer Portal
2. Create a new application
3. Go to "Bot" section and create a bot
4. Enable "Message Content Intent" under Privileged Gateway Intents
5. Copy the bot token (you'll add it to the config file in the next step)
6. Go to "OAuth2" > "URL Generator"
7. Select scopes: bot
8. Select permissions: Send Messages, Create Public Threads, Send Messages in Threads, Read Message History
9. Use the generated URL to invite the bot to your server

LLM Provider Setup

The bot needs an LLM API to function. Get an API key from one of these providers:

• OpenRouter: https://openrouter.ai/settings/keys (recommended - supports multiple models including Claude)
• OpenAI: https://platform.openai.com/api-keys

Any OpenAI-compatible API works. You can also use local LLMs with tools like Ollama or LM Studio.

Edit Configuration

Now edit config.yaml with your credentials:

1. Set bot.token to your Discord bot token (from Discord Bot Setup)
2. Set model.api_key to your LLM provider API key (from LLM Provider Setup)
3. Optionally customize model.base_url and model.name for your provider

See the comprehensive comments in config.yaml for all available options and examples.

Usage

uvx discord-llms --config config.yaml

Deploy (Docker)

Build and export an image with your config baked in:

make export CONFIG=path/to/config.yaml TAG=latest

You can also pass the config path positionally:

make export path/to/config.yaml TAG=latest

This creates discord-llms-<tag>.tar. Treat it as sensitive because it contains your secrets.

Load the image:

docker load -i discord-llms-latest.tar

Run it (logs persisted host):

docker run -d --name discord-llms \
  --restart unless-stopped \
  -v /volume1/docker/discord-llms/logs:/app/logs \
  discord-llms:latest