Summary

This is a discord bot written in C# which allows guilds to create a personality that they can interact with. The discord bot takes the user's message and sends it to Open AI's v1/chat/completions endpoint (docs here). The bot will prepend the request with the most recent message history from the current channel (default 20 messages), as well as a system prompt that defines the character that GPT4 will roleplay as. This prompt is assembled based on a list of "facts" that the guild may configure using /facts commands like /facts add.

An example of facts:

"You love wrestling, especially the attitude era."
"You are a fan of Star Trek."
"Your favorite food is german food. Especially sausages."
"You are an expert on EverQuest."

The above facts are the default for new guilds but can be managed by the guild through /facts commands.

Setup

This repo is currently designed to use GPT-4, as GPT-4 adheres to system prompts much much better than GPT-3.5. However, you might not immediately be granted access to GPT-4. If this is the case you can set the environment variable OpenAiOptions__Model=gpt-3.5-turbo or any other available model value. You may need to re-test the default prompts after making this change. Set this environment variable in the docker-compose.yml file at the root of the repo.

Included Third-Party Libraries

• Discord.NET
  • For Discord API and Socket interaction
• Polly
  • For managing retries with the HTTP Client for Open AI API
• LiteDB
  • Very lightweight database for storing per-guild configuration
  • By default this will be created in 2 places:
    • For Docker, in a docker volume called 'litedb' on the host machine
    • For local development on Windows, in a folder inside ProgramData/LiteDB
    • Location is driven by the configuration key DataServiceOptions__DatabasePath

Prerequisites

1. A Docker installation on your dev machine.
2. A Discord Bot Token
3. An Open AI API Token with GPT-4 access

Setup Steps

1. Clone the latest version of this repo
2. At the root, create a .env file with the following contents:

   BASE_IMAGE_NAME=your-base-image # e.g. myrepo/chatgptbot
   VERSION=1.0.0 # used as the docker image's tag
   DISCORD_TOKEN=your-discord-token
   OPENAI_API_KEY=your-openai-api-token
   SALT=your-custom-salt # this can be any random string, used to disguise the int ids for your underlying data

3. In /Daemon create a local.settings.json file with the following template:

   {
       "Secrets": {
           "DiscordToken": "your-discord-token",
           "OpenAiApiKey": "your-open-ai-api-key",
           "Salt": "your-custom-salt"
       }
   }

   The above template includes required values, but you can also modify other options. Check the classes located in /Options for other configurable settings. When debugging locally, values are read from local.settings.json. When running in docker, values are read from environment variables.

4. From a terminal at the root of the project, run docker compose with docker compose up -d
5. If setup was successful you should now have a container named gptbot running in docker.

Architecture Notes

The main source code is located in /Daemon.

Modifying the existing code is easy thanks to the simple project structure:

• /Accessors
  • Classes for accessing external services / data stores, including the OpenAI API and LiteDB database.
• /Builders
  • Builder classes for easily constructing prompts and ChatGPT messages.
• /Modules
  • Discord.NET Interaction Modules for /commands
• /Orchestrators
  • Top level orchestrators that manage the bulk of the code execution.