TAU-HOME.COM
LOADING

Scenario Open-Source '3D OS' Guide: Automating 3D Pipelines and Local DCC Control with 80 Agent Skills

An overview of Scenario's MIT-licensed open-source Agent Skills repository '3D OS' (scenario-labs/skills). Equip Claude Code, Cursor, and over 70 AI coding agen

tau · November 14, 2026

#Scenario #AgentSkills #3D-Modeling #ClaudeCode #GameDev

Scenario Open-Source '3D OS' Guide: Automating 3D Pipelines and Local DCC Control with 80 Agent Skills

Scenario's MIT-licensed open-source repository scenario-labs/skills—introduced publicly as "3D OS"—is a collection of Agent Skills that equips AI coding agents with end-to-end 3D production workflows and local digital content creation (DCC) software controls. Originally created on August 12, 2026, and actively maintained through release skills-v0.62.0 on November 11, 2026, the project was highlighted on Scenario's official X account on November 3, 2026, as a 6-department, 80-skill studio architecture for coding agents.

Architecture diagram of Scenario 3D OS Agent Skills showing six 3D production departments connected to local DCC tools and game engines

Image source: https://x.com/Scenario_gg/status/2117273548217524720

While modern coding agents excel at writing code and handling text, multi-stage 3D production requires passing assets across specialized disciplines—from concept art and modeling to rigging, texturing, and engine export. Rather than relying on a single monolithic prompt, Scenario structures this workflow as modular skills modeled after a game studio's department hierarchy.

Six Production Departments and the Expert Tools Layer

The repository organizes its workflow into six departments that hand off work sequentially from initial sketch to game-ready asset, paired with an Expert Tools tier for driving local software.

  • Animation: Covers character rigging (Rig Any Character) and video-to-motion conversion (Video to Animation).
  • Art: Handles storyboards and previsualization (Storyboards and Previs) alongside creature and character design (Creature and Character Design).
  • Modeling: Provides image-to-3D generation (Image to 3D Model) and hard-surface asset creation (Hard-Surface Vehicles and Weapons).
  • Look Dev and Lighting: Manages retexturing and PBR materials (Retexturing and PBR Materials) as well as rendering and lighting setups (Rendering and Lighting).
  • Worlds: Focuses on explorable 3D environments (Explorable 3D Worlds) and Gaussian splat scenes or skyboxes (Splat Scenes and Skyboxes).
  • Pipeline: Completes technical preparation with retopology, UVs, and baking (Retopology, UVs and Baking) plus engine-ready export (Engine-Ready Export).

Beneath these departments, the Expert Tools layer directly controls DCC applications and game engines installed on the user's local machine, supporting Blender, Maya, ZBrush, Unreal Engine, Unity, Godot, and Roblox Studio. According to Scenario's official clarification, local DCC and engine control skills—such as the Blender skill family—run locally using only the installed application without requiring a Scenario API key, whereas generative tasks like Image-to-3D route through the Scenario MCP server. The rigging workflow also includes a mechanical mode that binds rigid modular parts directly to single bones without weight painting.

Installation and MCP Setup

Built on the open Agent Skills specification (agentskills.io), the repository can be installed into more than 70 compatible coding agents—including Claude Code, Codex, Cursor, and Copilot—via the npx skills add CLI.

# Open the interactive installer menu to pick topic groups or individual skills
npx skills add scenario-labs/skills

# Install the core Scenario skill only
npx skills add scenario-labs/skills --skill scenario

To pin a reviewed release state instead of pulling from main, teams can lock both the CLI version and commit hash using npx [email protected] add scenario-labs/skills#<commit> --skill <name>. Because Expert Tools specialist skills import scripts from their family lead skill, each DCC or game engine family should be installed as a complete group via the installer menu or the README instructions inside skills/dcc/ and skills/game-engines/.

For generative skills backed by Scenario's cloud models, connect the Scenario MCP server ([https://mcp.scenario.com/mcp](https://mcp.scenario.com/mcp)) to your agent client and authenticate with a Scenario account (OAuth) or API key:

claude mcp add --transport http scenario https://mcp.scenario.com/mcp

Who It Fits and Practical Requirements

scenario-labs/skills is aimed at game developers, indie studios, and technical artists looking to integrate 3D asset production into agent-driven coding environments. Separating local DCC automation from cloud generative skills makes it practical to adopt the local Blender, Maya, or engine workflows immediately without incurring cloud generation costs.

Two prerequisites apply before deploying the skills in production:

  1. Pre-installed Local Applications: Expert Tools skills require Blender, Maya, ZBrush, Unreal Engine, Unity, Godot, or Roblox Studio to be installed locally on the host machine.
  2. Scenario Authentication for Generation: Generative workflows such as Image-to-3D require connecting the Scenario MCP server and authenticating with a valid Scenario account or API key.

Sources

Impact-Site-Verification: 9b3ebd10-6c45-44cf-a014-5b8c19b7b285