the-delegation  by arturitu

No-code 3D playground for agentic AI systems

Created 3 months ago
334 stars

Top 82.2% on SourcePulse

GitHubView on GitHub
Project Summary

A no-code, 3D interactive playground for designing and exploring agentic AI systems, "The Delegation" targets AI enthusiasts, educators, and developers. It simplifies understanding multi-agent collaboration through a visual interface and multimodal AI generation, eliminating the need for coding.

How It Works

This project employs a hybrid GPU/CPU architecture powered by Three.js (WebGPU) to simulate autonomous LLM agents within a shared 3D office environment. Agents navigate using intelligent pathfinding (three-pathfinding) and dynamic state machines, interacting via a node-based Team Editor (React Flow). The system integrates with the Gemini API for text, image, music, and video generation, allowing per-agent model assignment and real-time cost tracking.

Quick Start & Requirements

  • Install: npm install
  • Run: npm run dev
  • Access: Navigate to the local URL provided by the development server (typically http://localhost:3000/the-delegation).
  • Prerequisites: A Gemini API key is mandatory (BYOK - Bring Your Own Key).
  • Documentation: Links to "Launch The Delegation · Full Experience" and specific AI model prompt guides are mentioned.

Highlighted Details

  • Agentic AI System (v0.2.0): Features a visual Team Editor (React Flow) and 6 industry-specific predefined teams.
  • Multimodal Outputs: Supports generation of text, images (Nano Banana), music (Lyria 3), and video (Veo 3.1).
  • Embodied Simulation: Utilizes Three.js WebGPU for advanced rendering, with intelligent pathfinding and dynamic state machines for NPC behavior in a 3D workspace.
  • Workflow & Transparency: Includes PR-style workflows with human-in-the-loop approvals, cost/token tracking, and detailed technical logs.

Maintenance & Community

Developed by Arturo Paracuellos, with assets copyrighted by unboring.net. No explicit community channels or major contributor/sponsorship details are provided in the README.

Licensing & Compatibility

The project utilizes a dual-licensing model: source code is distributed under the permissive MIT License. However, custom 3D models and assets are licensed under CC BY-NC 4.0, prohibiting commercial use without explicit permission.

Limitations & Caveats

Operation is contingent on providing a personal Gemini API key. The CC BY-NC 4.0 license on 3D assets restricts commercial applications. The project is currently at version v0.2.0, suggesting ongoing development and potential for changes.

Health Check
Last Commit

1 month ago

Responsiveness

Inactive

Pull Requests (30d)
1
Issues (30d)
2
Star History
166 stars in the last 30 days

Explore Similar Projects

Starred by Eric Zhu Eric Zhu(Coauthor of AutoGen; Research Scientist at Microsoft Research), Luis Capelo Luis Capelo(Cofounder of Lightning AI), and
1 more.

hyperframes by heygen-com

7.8%
21k
HTML-native video rendering for AI agents
Created 2 months ago
Updated 13 hours ago
Feedback? Help us improve.