eliautobot

my-virtual-world: Host a 3D World for AI Agents

My Virtual World is a self-hosted 3D environment where AI agents can move, interact with objects, and show activity from local agent systems. It suits people who want a visual workspace for OpenClaw, Hermes, or other agent harnesses.

JavaScriptOther licenseAI AgentsSelf HostedMulti Agent
View on GitHub Compare
Stars
80
Forks
12
Stars, last 29 days
+6

Our take

Health score

30/100

How it is scored
Activity
74
Community
4
Issues
21
Pull requests
21

+6 stars in the last 29 days

  • Releases more often than 89% of the projects we track
  • Fewer contributors than 91% of the projects we track

Project history

Daily GitHub snapshots since Oct 7, 2026. History covers stars and forks; health signals are not tracked historically.

Selected snapshot

Stars

87

Forks

13

Change since Nov 4, 2026: 0 stars·0 forks

Overview

My Virtual World turns local AI agent activity into a navigable 3D space. Agents can occupy buildings, move between areas, use objects, and appear alongside live activity from connected systems such as OpenClaw and Hermes.

It is aimed at people running agent harnesses on their own machines or private networks who want a shared, visual interface for agent presence and interactions. It is less suitable if you only need a text-based agent interface or want an app exposed publicly without additional security controls.

Key Features

  • Renders a voxel-style world with Three.js, including roads, buildings, interiors, and outdoor spaces.
  • Supports agent movement, seating, object use, queues, and world actions.
  • Integrates optionally with OpenClaw for live agent presence and chat activity.
  • Supports local Hermes profiles and includes Codex workspace agent support.
  • Stores persistent world data as JSON.
  • Provides optional Agent Browser and SMS/Twilio integrations when licensed and configured.
  • Offers Docker-first deployment and a local development path.

Use Cases

  • OpenClaw users can visualize agent presence and activity while their local gateway is running.
  • Hermes users can connect local profiles and represent agent work in a persistent world.
  • Teams or individuals building multi-agent workflows can use buildings and objects as a spatial interface for agent activity.
  • Developers can run the project on a local machine or private network to explore agent-world interactions without relying on a public hosted service.

Getting Started

The recommended Docker setup is:

git clone https://github.com/eliautobot/my-virtual-world.git
cd my-virtual-world
cp .env.example .env
docker compose up --build -d

Open http://localhost:8590, then visit http://localhost:8590/setup. See the README and installation guide for configuration and integration details.

Considerations

  • The README identifies the project as source-available, not OSI-approved open source. It describes the software license as PolyForm Noncommercial 1.0.0, with commercial use requiring a separate written license. The repository metadata reports NOASSERTION.
  • Demo mode limits the starter world and keeps advanced editing, Agent Browser, SMS/Twilio, and Agent Live Mode locked until license activation.
  • Docker is recommended and requires Docker plus at least 4 GB of free RAM and 2 GB of free disk space. Local development requires Python 3.12+ and Node.js 22+.
  • OpenClaw, Hermes, and other integrations are optional, and depend on the corresponding local systems being available and configured.
  • The project is intended for trusted machines or private networks. The README warns against exposing its app, realtime runtime, or agent service ports directly to the public internet without authentication and network hardening.

Found this useful?

Share it with someone who would like my-virtual-world.

Comparisons