0xAI

0xAI is a Fabric client mod that adds an Ollama-powered chat companion to Minecraft. Type @ai in chat and the mod sends the prompt to your local Ollama server, then replies in chat.

Support

• Mod version: 0.1.5
• Minecraft: 26.1.x
• Fabric Loader: 0.17.3+
• Fabric API: required
• Java: 25+
• Mod Menu: optional, supported

Mod Menu adds an in-game config screen for the same settings stored in config/oxai.json.

Install

1. Install Fabric Loader and Fabric API for Minecraft 26.1.x.
2. Put the 0xAI JAR in your mods folder.
3. Optional: install Mod Menu to configure 0xAI in-game.
4. Start Ollama:

ollama pull llama3.2 ollama serve

1. Launch Minecraft.

Usage

In chat:

@ai how do i find diamonds @ai what's the best food @ai how do i make a nether portal

The default trigger is @ai. You can change it in Mod Menu or in config/oxai.json.

Configuration

With Mod Menu installed:

1. Open Mods.
2. Select 0xAI.
3. Open the config screen.
4. Change settings and press Done.

Without Mod Menu, edit config/oxai.json:

{ "enabled": true, "trigger": "@ai", "ollamaUrl": "http://localhost:11434", "model": "llama3.2:3b", "systemPrompt": "You are a helpful AI assistant in a Minecraft server. Keep responses concise and friendly.", "maxTokens": 150, "temperature": 0.7, "responsePrefix": "[AI] ", "maxMessageLength": 256, "showErrors": true, "requestTimeout": 60000, "contextMessages": 5 } Option What it does enabled Turns AI replies on or off trigger Chat text that triggers the AI ollamaUrl Ollama server address model Ollama model name systemPrompt System message sent with each request maxTokens Maximum generated response length temperature Response randomness from 0.0 to 1.0 responsePrefix Prefix added before AI chat messages maxMessageLength Maximum characters per chat line showErrors Shows request errors in chat requestTimeout Ollama request timeout in milliseconds contextMessages Number of recent chat turns to keep as context

Models

0xAI works with any Ollama chat model. Common choices:

• llama3.2:3b
• llama3.2:1b
• mistral
• phi3

Use smaller models for faster replies or lower-end hardware.

Build

Install JDK 25+, then build with Gradle:

./gradlew build

On Windows PowerShell, if the Unix wrapper script is not executable, run:

java -classpath gradle\wrapper\gradle-wrapper.jar org.gradle.wrapper.GradleWrapperMain build

The built JAR is written to build/libs/.

Troubleshooting

Build says release version 25 not supported

Gradle is using an older JDK. Install JDK 25 and make sure JAVA_HOME or your IDE Gradle JVM points to it.

The mod does not respond

• Make sure 0xAI is enabled in Mod Menu or config/oxai.json.
• Check that Ollama is running with ollama serve.
• Check that the configured model is installed with ollama list.
• Verify that your message contains the configured trigger.

Replies are slow

• Use a smaller model such as llama3.2:1b.
• Lower maxTokens.
• Lower contextMessages.

Connection refused

• Check ollamaUrl.
• Make sure Ollama is running on the configured host and port.