localserverassistantai

LocalServerAssistantAI

A local AI-powered Markdown reference-type server guide + free chat + Discord integration plugin for Paper 26.2+ Minecraft servers.

Uses Ollama or OpenAI-compatible local APIs to perform RAG search on .md files placed in plugins/LocalServerAssistantAI/Documents/.

This project was inspired by the desire to replicate the functionality of the paid cloud AI plugin ServerAssistantAI using local AI.

This project was created using AI and is inspired by Tribie's project on X.

Features

• RAG Search: Splits Markdown documents into sections by heading, indexes them, and passes only relevant parts to the AI
• Hybrid Japanese n-gram + Embedding Search: Combines lexical and semantic search for high-precision document retrieval
• Server Query Mode (/sai ask / /sai mode docs): Generates answers based only on reference documents
• Free Chat Mode (/sai chat / /sai mode chat): Conversational AI with maintained history
• Chat Hijack Mode: Enable with /sai mode to redirect all chat input to the AI
• Discord Integration: Configure separate channels for questions and free chat
• Ollama / OpenAI-Compatible Local API support (LM Studio / LocalAI / llama.cpp server)
• Rate Limiting, Concurrency Control, History Clearing: Server load and spam protection
• Fully Asynchronous: Does not block Paper's main thread while waiting for AI responses

Requirements

• Paper 26.2+ (Minecraft 1.21.4)
• Java 25+
• Local AI Server (Recommended: Ollama)
• Discord Bot (only if using Discord integration)

Build

build.bat

./build.sh

On first run, Gradle 9.6.1 and Java 25 toolchain will be downloaded automatically. Output: build/libs/LocalServerAssistantAI-1.0.1.jar

Ollama Setup

Install Ollama

Windows: Download the installer from ollama.com and run it.

Linux: ```bash curl -fsSL https://ollama.com/install.sh | sh

**macOS:**
Download the macOS app from [ollama.com](https://ollama.com) or use Homebrew:
```bash
brew install ollama

Pull Models

ollama pull qwen3:8b
ollama pull embeddinggemma

Configuration

Edit plugins/LocalServerAssistantAI/config.yml.

ai:
  provider: ollama
  ollama:
    base-url: "http://127.0.0.1:11434"
    chat-model: "qwen3:8b"
    embedding-model: "embeddinggemma"

Environment variables can also be used:

• LOCAL_SERVER_ASSISTANT_AI_API_KEY - API key for OpenAI-compatible APIs (was LOCAL_SERVER_ASSISTANT_CLONE_API_KEY before v1.0.1)
• LOCAL_SERVER_ASSISTANT_AI_DISCORD_TOKEN - Discord Bot token (was LOCAL_SERVER_ASSISTANT_CLONE_DISCORD_TOKEN before v1.0.1)

Usage (Minecraft)

Permissions

Note: From v1.0.1, the .clone. part has been removed from permission node names.

Discord Integration

Configure the discord section in config.yml.

discord:
  enabled: true
  token: "BOT_TOKEN"
  guild-id: "DISCORD_SERVER_ID"
  ask-channel-ids: ["CHANNEL_ID"]
  chat-channel-ids: ["CHANNEL_ID"]
  command-prefix: "!ai"
  answer-all-messages-in-configured-channels: true

In ask channels, use !ai ask <question>. In chat channels, use !ai chat <message>.

Example Documents Structure

Documents/
├── rules.md
├── commands.md
└── worlds/
    └── resource-world.md

Run /sai reindex after making changes.

License

MIT License