Installation_steps.md

MinePal - AI-Agent with RL capabilites in a Minecraft world - Installation

A step-by-step guide to creating an offline autonomous Minecraft AI bot, powered by a local LLM (LM Studio Meta‑LLaMA‑3.1‑Instruct), playing alongside you on a Fabric 1.21 server managed with TLauncher.

Prerequisites

• Java 17+ installed (required for Fabric and Minecraft ≥ 1.18) :contentReference[oaicite:1]{index=1}
• Node.js installed
• TLauncher with Fabric 1.21 profile
• LM Studio v0.3.6+ with local API enabled
• Official Fabric Server installer (universal JAR)

Step 1: Set Up Java

Install Java 17+ and verify with:

java -version
# Expect something like "openjdk version \"17.x.x\""

Minecraft ≥ 1.18 and Fabric ≥ 1.17 require Java 17+ ([wiki.fabricmc.net][1], [github.com][2]).

Step 2: Install Fabric Server (1.21)

1. Download the Fabric Installer JAR from the Fabric website.

2. In an empty folder, run:

   java -jar fabric-installer.jar

   • Select Server

   • Choose Minecraft 1.21.x

   • Click Install, generating:

     • fabric-server-launch.jar
     • start.sh/start.bat
     • Dependencies including server.jar ([wiki.fabricmc.net][3], [wiki.fabricmc.net][1])

Step 3: Configure Offline Server

In the server folder, update:

• eula.txt → eula=true

• server.properties →

  online-mode=false
  enable-command-block=true

These settings allow TLauncher (cracked) players to connect ([wiki.fabricmc.net][1], [wiki.fabricmc.net][3]).

Step 4: Start Fabric Server

• On Windows: run start.bat

• On Linux/macOS:

  chmod +x start.sh
  ./start.sh

The console should show the server is online on port 25565.

Step 5: Join with TLauncher

1. Open TLauncher with Fabric 1.21 profile.

2. Navigate to Multiplayer → Direct Connect

3. Enter:

   localhost
   

You should now join your offline Fabric server world.

Step 6: Set Up LM Studio Local API

1. Install LM Studio v0.3.6+.

2. In the Developer tab, enable the OpenAI‑compatible HTTP API at http://localhost:1234/v1/…

3. Test with:

   curl http://localhost:1234/v1/models

   You should see "meta-llama-3.1-8b-instruct:2" listed ([youtube.com][4], [wiki.fabricmc.net][3], [wiki.fabricmc.net][1], [lmstudio.ai][5]).

4. Test tool-calling:

   curl -i http://localhost:1234/v1/chat/completions \
     -H "Content-Type: application/json" \
     -d '{
       "model":"meta-llama-3.1-8b-instruct:2",
       "messages":[{"role":"system","content":"You can call functions."},
                   {"role":"user","content":"Call echo with text \"hello world\""}],
       "functions":[{
         "name":"echo",
         "description":"Returns the same text",
         "parameters":{"type":"object","properties":{"text":{"type":"string"}},"required":["text"]}
       }]
     }'

   ✅ Confirm response has tool_calls JSON .

Step 7: Create the Bot Project

mkdir mc-ai-bot
cd mc-ai-bot
npm init -y
npm install mineflayer mineflayer-pathfinder mineflayer-auto-eat mineflayer-collectblock node-fetch minecraft-data

Save the complete bot.js script (provided in the code block below).

Step 8: bot.js Overview

• Connects to localhost:25565 and identifies as LLM_Bot

• Defines tools mineBlock and followPlayer for LM to call

• Handles chat:

  • Parses tool_calls JSON and runs in-game actions
  • Cleans markdown, supports alternative action JSON format

• Sends feedback to LLM to maintain context

(See code in bot.js)

Step 9: Launch the Bot

Ensure both server and TLauncher are running, then in your bot directory:

node bot.js

You should see in-game chat:

LLM_Bot: Hello! I'm your AI teammate. Chat with me!

Step 10: Test Commands

In the Minecraft chat, type:

• @LLM_Bot please mine oak_log
• @LLM_Bot follow me

Your bot should mine logs and follow you accordingly!

Step 11: Extend Functionality

Add more tools:

• storeItems, eatFood, fightMob, etc.
• Provide contextual feedback (inventory, health).
• Refine prompts or use function-calling for behavior chaining.

Troubleshooting

| Issue | Solution | | - | | | Fabric server crashes | Ensure Java 17+ is used ([docs.mcserversoft.com][6], [wiki.fabricmc.net][1], [github.com][7], [youtube.com][8]) | | Bot fails to parse JSON | Use cleanJson() to strip markdown | | Function calls not triggered | Confirm LM Studio v0.3.6+ supports OpenAI-style API |

📂 bot.js Code

Use the one which is present in this repository

Conclusion

You’ve set up a fully offline, autonomous AI companion using TLauncher + Fabric + LM Studio + Mineflayer, complete with structured tool-calling for in-game behaviors. Ready to play and extend!

Happy coding & gaming! ⚔️🚀