Spitball

Minecraft 版本： 1.21.4

平台： fabric

标签： management social

What is Spitball?

Spitball is a powerful server-side mod that gives your players access to an AI assistant you control completely. With Spitball, the AI can be tailored precisely to your server’s needs. Whether it's answering FAQs or guiding players through custom game mechanics, Spitball serves as an in-game help desk without the need for human moderators.

What can it do?

Replace the need for help from staff by providing accurate, on-demand answers to player questions.

• Utilize your own server-specific data (via the additionalInfo setting) to deliver context-aware answers.

• From game rules to in-game tips, the AI offers useful insights without executing commands or interfering with gameplay.

  What can't it do?

• The AI is strictly advisory. It won’t run commands or alter game states.

• There’s no Minecraft-specific content moderation built into Spitball, it depends on your system prompt and model

Just how configurable is it?

Currently, everything that can be configured is located in the spitball.toml config file. That file - by default - looks as follows;

# Configuration for Spitball

# Permission level (between 0 and 4). Default: 0
permissionLevel = 0

# Whether to broadcast messages to ops. Default: false
broadcastToOps = false

# The host URL for Ollama. Default: empty
ollamaHostURL = ""

# Maximum requests per minute. Default: 1
maxRequestsPerMinute = 1

# Whether ops bypass the request limit. Default: true
opsBypassLimit = true

# System prompt. Default: You are a helpful AI on a minecraft server
systemPrompt = "You are a helpful AI on a minecraft server"

# Additional info about your Server the AI should have. Default: empty
additionalInfo = ""

# The model on your ollama instance to be used. Default: empty
ollamaModel = ""

# The command used to access the AI. Default: "guide"
chatCommand = "guide"

# The prefix AI responses have in your chat. Default: "[Guide]"
responsePrefix = "§c[§bGuide§c]§r"

# Whether the AI should use chat formatting (Additional information sent to AI, so response times are increased). Default: false
chatFormatting = false