Pendulum

Client-side JavaScript executor for Minecraft. Automate player actions — movement, construction, combat, inventory,
world queries, and Baritone supported. Designed for both human players and AI agents.

This mod is partly inspired by Playwright.js, which allows agents to control web browsers via JavaScript. Pendulum
brings that same power to Minecraft, enabling complex automation and AI integration through a simple JS API.

Version Support

| Minecraft | Fabric | Forge | NeoForge | Status |
|-----------|--------|-------|----------|-----------------|
| 1.20.1 | ✅ | ✅ | | Fully Supported |
| 1.21.1 | ✅ | ❌ | ✅ | Fully Supported |
| 26.1.2 | 🔨 | ❌ | 🔨 | In development |

Features

For detailed documentation, see the Documentation

🧑 For Players

• /pendulum execute — run JavaScript inline in chat
• /pendulum file — run .js scripts from .minecraft/pendulum/
• mc.* — 80+ API functions: mc.player.forward(), mc.world.findBlocks(), mc.inv.hasItem(), mc.gui.click(), and more. Log via pendulum.log(), pendulum.warn(), pendulum.error().
• br.* — 45 Baritone functions: pathfinding, mining, farming, building schematics

🤖 For AI Agents

• MCP JSON-RPC 2.0 over TCP — native AI agent integration
• script/eval — execute arbitrary JS and return results
• script/status / script/abort — manage script state
• gui/enumerateWidgets — full recursive widget tree
• gui/guiElements — read all visible GUI controls
• gui/screenshot — capture the game view
• simulate/click / gui/clickButton — click GUI elements
• simulate/typeText / simulate/pressKey — keyboard input
• Compatible with VS Code Copilot, Claude Desktop, and any MCP client

Quick Start

// Walk forward 1 second
/pendulum execute mc.player.forward(20)

// Mine all pumpkins nearby
/pendulum execute for(let p of mc.world.findBlocks('minecraft:pumpkin',8)){ mc.player.breakBlockAt(p.x,p.y,p.z) }

// Run a script file
/pendulum file farm.js

// AI Agent workflow: screenshot + click
gui/screenshot  →  gui/clickButton("Done")  →  simulate/typeText("Hello", true)

MCP Server

Start the MCP server for AI agent access:

/pendulum mcp start      # default port 25566
/pendulum mcp stop       # stop the server
/pendulum mcp status     # check if running

Configure your MCP client to connect to localhost:25566 (default).