← Open Source
arturitu

the-delegation

A no-code 3D playground to explore, design, and interact with Agentic AI systems.

AI EngineeringMake agent see hear talkBuild agent workflow (visual)TypeScript
Open on GitHub
Momentum
+4stars in 24 hours+0.6%
698
Stars
131
Forks
+19
This week
1
Contributors
Created 2026-02-19 · Updated 2026-10-16 · #2608 today
Top developers

arturitu/the-delegation is ranked 2,608 with 698 stars.

Same-category comparison

arturitu/the-delegation is ranked 2,608 among these 8 peers; aiworkskills/wechat-article-skills is ranked 2,609.

Same-category comparison
NameRankMetrics
gethopp/figma-mcp-bridge2,598
Language:
TypeScript
Stars:
757
Forks:
105
Description:
Figma Plugin & MCP server to bypass API limits.
HelpCode-ai/anythingmcp2,599
Language:
TypeScript
Stars:
751
Forks:
87
Description:
Open-source, self-hosted MCP gateway: turn any REST/OpenAPI, SOAP, GraphQL, OData or SQL API into MCP tools for Claude, ChatGPT & Copilot. 474 connectors incl. SAP S/4HANA & Business One, ERP, e-commerce.
slee-persis/GVS5H2,600
Language:
Python
Stars:
750
Forks:
91
Description:
GVS5H: Five Qwen3.8-27B Models Match Claude Fable 5 on LiveCodeBench Hard | Fable 5 Level Coding for a Fifth the Price - or on a Single GPU.
harbor-framework/terminal-bench-science2,603
Language:
Python
Stars:
722
Forks:
477
Description:
Terminal-Bench-Science: Evaluating AI agents on research workflows across scientific domains.
aiworkskills/wechat-article-skills2,609
Language:
Python
Stars:
692
Forks:
104
robocurve/inspect-robots2,611
Language:
Python
Stars:
682
Forks:
84
Description:
Open source evals for physical AI. Run any LLM/VLA on any arm/humanoid against any real/sim benchmark.
JayPokale/Chisle2,612
Language:
JavaScript
Stars:
673
Forks:
49
Description:
Cut your AI coding agent's token bill on three axes: terse prose, YAGNI-first code, and tool-output compression. Claude Code, Pi, Cursor, Codex, Gemini + 4 more. Zero deps, published benchmarks including the runs it loses.
agentenatalie/get-job.skill2,614
Language:
Python
Stars:
654
Forks:
19
Key parameters and specifications
Key parameters and specifications
ParameterValue
LanguageTypeScript
CategoryAI Engineering
SubcategoryMake agent see hear talk, Build agent workflow (visual)
KeywordsAgent sandbox, AI character, Multimodal, No code/Low code
Stars698
Forks131
Contributors1
Created date2026-02-19
Updated date2026-10-16
Stars in 1 day4
Stars in 7 days19
Rank2,608
Previous rank4,337
Rank change1,729
7-day stars rank2,843
Previous 7-day stars rank3,411
7-day stars rank change568
Total stars rank14,915
Previous total stars rank14,945
Total stars rank change30
README summary

Launch The Delegation · Full Experience * [! IMPORTANT] **\*** This experience requires **BYOK (Bring Your Own Key)**. You will need a **Gemini API key** to run the simulation. Deep integration enables native support for text, and multimodal generation (**Nano Banana**, **Lyria 3**, **Veo 3. You can also **clone or fork** this repository to run it locally. A no-code 3D playground to explore, design, and interact with Agentic AI systems This project is designed for **AI enthusiasts, educators, and creative developers** looking to

understand multi-agent collaboration in a living 3D office without writing a single line of code. **Open the app:** Navigate to the local URL shown in your terminal (usually ). 0) **Team Editor (React Flow):** Create your own multi-agent design patterns using an interactive node-based interface. **6 Predefined Teams:** Industry-specific templates (Creative Agency, Film Studio, PR Agency, etc. **Multimodal Outputs:** Generate professional assets including text, image (Nano Banana), music (Lyria 3), and video (Veo 3. **Per-Agent LLM

:** Assign different Gemini Models to specific roles (e. **Cost & Token Tracking:** Real-time estimation of usage costs and token consumption for transparency. **PR-style Workflows:** Learn about Pull Request and Review workflows where agents with properties require your approval to proceed. **Guardrails:** Controlled generation with the option, ensuring quality before final asset production. **Technical Logs:** Improved visibility into raw LLM traces, tool calls, and structured agent responses. Embodied Simulation **Hybrid GPU/CPU Architecture:** A high-performance **3D simulation** built with

**Three. js WebGPU** where autonomous LLM-powered characters collaborate in a shared physical workspace. **Intelligent Pathfinding:** NPCs utilize a NavMesh to navigate the office, finding and claiming specific "Points of Interest" (desks, seats, computers) based on their current task. Pathfinding is powered by three-pathfinding. **Dynamic State Machine:** Characters transition naturally between walking, sitting, working, and talking, with sync'ed 3D speech bubbles and expressions. Interactive UI **Team Flow Visualizer:** Real-time node-based view of your agent hierarchy and task

flows. **Simulated PR Reviews:** Interactive modals for reviewing agent proposals, providing feedback, and merging tasks. **Real-time 3D Overlay:** Status indicators and interaction menus projected from 3D space into a polished UI. **Agent Inspector:** Select any agent to view their "thoughts", mission, and history. **Kanban & Action Logs:** Complete transparency into the agency's progress and tool-level interactions. js (WebGPU & TSL) for advanced rendering and compute. **UI:** React & React Flow for node-based team visualization. **AI:** Gemini API is the core

LLM provider. We follow official prompting best practices for: **State:** Zustand for a unified, reactive store across the 3D world and React UI. **3D Assets:** Custom models and animations rigged in Blender, using an instanced animation system. Roadmap **World Building** **Advanced Interactions** **Refinement** Developer Note This release (**v0. 0**) was developed entirely using **Google Antigravity** as the primary IDE and powered by **Gemini 3 Flash**. License & IP This project follows a dual-licensing model: **Source Code (MIT):** All logic, shaders

, and UI code are free to use, modify, and distribute. 0):** The custom 3D office and character models are Copyright

Frequently asked questions

How many stars and forks are recorded?

arturitu/the-delegation has 698 stars and 131 forks.

What is the repository rank?

arturitu/the-delegation is ranked 2,608.

What language and category are recorded?

arturitu/the-delegation has 698 stars, uses TypeScript and belongs to AI Engineering.

Which repository is closest in rank?

arturitu/the-delegation is compared with aiworkskills/wechat-article-skills, ranked 2,609 with 692 stars.

README

Launch The Delegation · Full Experience *

[!IMPORTANT] * This experience requires BYOK (Bring Your Own Key). You will need a Gemini API key to run the simulation. Deep integration enables native support for text, and multimodal generation (Nano Banana, Lyria 3, Veo 3.1). You can also clone or fork this repository to run it locally.

What is The Delegation?

A no-code 3D playground to explore, design, and interact with Agentic AI systems

This project is designed for AI enthusiasts, educators, and creative developers looking to understand multi-agent collaboration in a living 3D office without writing a single line of code.

Getting Started

  1. Install dependencies:
npm install
  1. Run the development server:
npm run dev
  1. Open the app: Navigate to the local URL shown in your terminal (usually http://localhost:3000/the-delegation).

Features

Agentic AI System (v0.2.0)

  • Team Editor (React Flow): Create your own multi-agent design patterns using an interactive node-based interface.
  • 6 Predefined Teams: Industry-specific templates (Creative Agency, Film Studio, PR Agency, etc.) to get you started.
  • Multimodal Outputs: Generate professional assets including text, image (Nano Banana), music (Lyria 3), and video (Veo 3.1) directly from your agent teams.
  • Per-Agent LLM: Assign different Gemini Models to specific roles (e.g., Flash for speed, Pro for reasoning).
  • Cost & Token Tracking: Real-time estimation of usage costs and token consumption for transparency.
  • PR-style Workflows: Learn about Pull Request and Review workflows where agents with human-in-the-loop properties require your approval to proceed.
  • Guardrails: Controlled generation with the Auto-approve output option, ensuring quality before final asset production.
  • Technical Logs: Improved visibility into raw LLM traces, tool calls, and structured agent responses.

Embodied Simulation

  • Hybrid GPU/CPU Architecture: A high-performance 3D simulation built with Three.js WebGPU where autonomous LLM-powered characters collaborate in a shared physical workspace.
  • Intelligent Pathfinding: NPCs utilize a NavMesh to navigate the office, finding and claiming specific "Points of Interest" (desks, seats, computers) based on their current task. Pathfinding is powered by three-pathfinding.
  • Dynamic State Machine: Characters transition naturally between walking, sitting, working, and talking, with sync'ed 3D speech bubbles and expressions.

Interactive UI

  • Team Flow Visualizer: Real-time node-based view of your agent hierarchy and task flows.
  • Simulated PR Reviews: Interactive modals for reviewing agent proposals, providing feedback, and merging tasks.
  • Real-time 3D Overlay: Status indicators and interaction menus projected from 3D space into a polished UI.
  • Agent Inspector: Select any agent to view their "thoughts", mission, and history.
  • Kanban & Action Logs: Complete transparency into the agency's progress and tool-level interactions.

Tech Stack Deep Dive

Roadmap

  • World Building
    • [ ] Office/3D Space Editor: Drag-and-drop workspace layout and POI customization.
    • [ ] Dynamic Environment: Real-time prop generation and environment modification by agents.
  • Advanced Interactions
    • [ ] Advanced Embodied AI: Deeper integration between agent reasoning and physical 3D world actions.
    • [ ] Enhanced Animations: Richer character expressions and more fluid, context-aware animations.
    • [ ] Human-Agent Spatial Interaction: Direct collaboration and richer multi-party interactions in the 3D office.
    • [ ] Inter-Agent Knowledge Sharing: Long-term memory for agent teams across projects.
  • Refinement
    • [ ] Architecture Decoupling: Further separation of core logic from the simulation environment.
    • [ ] UX/UI Overhaul: Unified CSS styling based on "The Delegation" brand identity.

Developer Note

This release (v0.2.0) was developed entirely using Google Antigravity as the primary IDE and powered by Gemini 3 Flash.

License & IP

This project follows a dual-licensing model:

  • Source Code (MIT): All logic, shaders, and UI code are free to use, modify, and distribute.
  • 3D Models & Assets (CC BY-NC 4.0): The custom 3D office and character models are Copyright © 2026 Arturo Paracuellos (unboring.net). They are free for personal and educational use but cannot be used for commercial purposes without permission.

Developed with ❤️ by Arturo Paracuellos

Nearby in the ranking
Similar items