Token导航 LogoToken导航TokenDH.com
研究检索需要联网github未标认证来源可访问许可证需确认审计提醒

add-3d-models添加 3d 模型

Agent Skill

add-3d-models 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

541

周安装

23

GitHub Stars

13

下载量

190
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:add-3d-models(添加 3d 模型)
来源仓库:https://github.com/dcl-regenesislabs/opendcl
仓库路径:skills/add-3d-models
安装命令:
npx skills add https://github.com/dcl-regenesislabs/opendcl --skill add-3d-models
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/dcl-regenesislabs/opendcl --skill add-3d-models

简介

add-3d-models 用于在 Decentraland 场景中加载 glb/gltf 格式的 3D 模型文件。

  • 支持设置模型位置、旋转、缩放及碰撞掩码,确保模型在虚拟世界中可点击和交互。
  • 通过 GltfContainer 组件加载模型,并建议将可见网格作为碰撞体以保证物理准确性。
  • 安装命令为 npx skills add https://github.com/dcl-regenesislabs/opendcl --skill add-3d-models。
  • 使用时需注意模型路径正确性及 visibleMeshesCollisionMask 的配置,避免穿透或不可交互问题。

SKILL.md

Adding 3D Models to Decentraland Scenes

Loading a 3D Model

Use GltfContainer to load .glb or .gltf files:

import { engine, Transform, GltfContainer, ColliderLayer } from '@dcl/sdk/ecs'
import { Vector3, Quaternion } from '@dcl/sdk/math'

const model = engine.addEntity()
Transform.create(model, {
  position: Vector3.create(8, 0, 8),
  rotation: Quaternion.fromEulerDegrees(0, 0, 0),
  scale: Vector3.create(1, 1, 1)
})
GltfContainer.create(model, {
  src: 'models/myModel.glb',
  visibleMeshesCollisionMask: ColliderLayer.CL_PHYSICS | ColliderLayer.CL_POINTER
})
Always set visibleMeshesCollisionMask when loading models. Catalog models don't include separate collider meshes — using the visible mesh as the collider ensures the model is solid and clickable.

File Organization

Place model files in a models/ directory at the project root:

project/
├── models/
│   ├── building.glb
│   ├── tree.glb
│   └── furniture/
│       ├── chair.glb
│       └── table.glb
├── src/
│   └── index.ts
└── scene.json

Colliders

Using Model's Built-in Colliders

Models exported with collision meshes work automatically. Set the collision mask:

GltfContainer.create(model, {
  src: 'models/building.glb',
  visibleMeshesCollisionMask: ColliderLayer.CL_PHYSICS | ColliderLayer.CL_POINTER,
  invisibleMeshesCollisionMask: ColliderLayer.CL_PHYSICS
})

Adding Simple Colliders

For basic shapes, add MeshCollider:

import { MeshCollider } from '@dcl/sdk/ecs'
MeshCollider.setBox(model) // Box collider
MeshCollider.setSphere(model) // Sphere collider

Common Model Operations

Scaling

Transform.create(model, {
  position: Vector3.create(8, 0, 8),
  scale: Vector3.create(2, 2, 2) // 2x size
})

Rotation

Transform.create(model, {
  position: Vector3.create(8, 0, 8),
  rotation: Quaternion.fromEulerDegrees(0, 90, 0) // Rotate 90° on Y axis
})

Parenting (Attach to Another Entity)

const parent = engine.addEntity()
Transform.create(parent, { position: Vector3.create(8, 0, 8) })

const child = engine.addEntity()
Transform.create(child, {
  position: Vector3.create(0, 2, 0), // 2m above parent
  parent: parent
})
GltfContainer.create(child, { src: 'models/hat.glb' })

Free 3D Models — OpenDCL Catalog (5,700+ models)

The catalog file is at {baseDir}/references/model-catalog.md. Each line has this format:

slug | dims | tris | size | category/sub | description [tags] [anim: clips] | curl command | preview: thumbnail_url

How to search

Search with one keyword at a time — try the most specific word first:

grep -i "zombie" {baseDir}/references/model-catalog.md

If no results, try synonyms, broader terms, or related words:

  • "sofa" → "couch" → "seat" → "furniture"
  • "car" → "vehicle" → "truck" → "van"
  • "wall" → "fence" → "barrier" → "structure"

Browse all categories to discover what's available:

grep "^##" {baseDir}/references/model-catalog.md

Search within a specific category:

grep "^##\|chair" {baseDir}/references/model-catalog.md

How to use models

  1. Search the catalog with different keywords until you find matching models
  2. Review the results — check dimensions, triangle count, animations, and description
  3. Download the model with the curl command into models/
  4. Reference in code with GltfContainer.create(entity, {src: 'models/{slug}.glb'})
  5. If the model has animations (listed in [anim:...]), use the Animator component to play them
  6. After placing the model, you can fetch its preview thumbnail (preview: URL) to see what it looks like

Example workflow

# Search for zombie models
grep -i "zombie" {baseDir}/references/model-catalog.md

# Found: zombie-purple | 2.8×2.9×0.5m | 1472 tri | 271KB | character/zombie | ...
#   [anim: Tpose, ZombieAttack, ZombieUP, ZombieWalk]
#   preview: https://models.dclregenesislabs.xyz/blobs/bafkrei...

# Download the model
curl -o models/zombie-purple.glb "https://models.dclregenesislabs.xyz/blobs/bafybeiffc..."
// Use in code with animations
import { engine, Transform, GltfContainer, Animator } from '@dcl/sdk/ecs'
import { Vector3 } from '@dcl/sdk/math'

const zombie = engine.addEntity()
Transform.create(zombie, { position: Vector3.create(8, 0, 8) })
GltfContainer.create(zombie, { src: 'models/zombie-purple.glb' })
Animator.create(zombie, {
  states: [
    { clip: 'ZombieWalk', playing: true, loop: true },
    { clip: 'ZombieAttack', playing: false, loop: false }
  ]
})
Important: GltfContainer only works with local files. Never use external URLs for the model src field. Always download models into models/ first. Never cd into the models directory. Always run curl from the project root with curl -o models/slug.glb "URL". Do NOT use cd models && curl -o slug.glb.

Checking Model Load State

Use GltfContainerLoadingState to check if a model has finished loading:

import { GltfContainer, GltfContainerLoadingState, LoadingState } from '@dcl/sdk/ecs'

engine.addSystem(() => {
  const state = GltfContainerLoadingState.getOrNull(modelEntity)
  if (state && state.currentState === LoadingState.FINISHED) {
    console.log('Model loaded successfully')
  } else if (state && state.currentState === LoadingState.FINISHED_WITH_ERROR) {
    console.log('Model failed to load')
  }
})

Troubleshooting

ProblemCauseSolution
Model not visibleWrong file pathVerify the file exists at the exact path relative to project root (e.g., models/myModel.glb)
Model not visiblePosition outside scene boundariesCheck Transform position is within 0-16 per parcel. Center of 1-parcel scene is (8, 0, 8)
Model not visibleScale is 0 or very smallCheck Transform.scale — default is (1,1,1). Try larger values if model was exported very small
Model not visibleBehind the cameraMove the avatar or rotate to look in the model's direction
Model loads but looks wrongY-up vs Z-up mismatchDecentraland uses Y-up. Re-export from Blender with "Y Up" checked
"FINISHED_WITH_ERROR" load stateCorrupted or unsupported.glbRe-export the model. Use .glb (binary GLTF) format. Ensure no unsupported extensions
Clicking model does nothingMissing colliderAdd visibleMeshesCollisionMask: ColliderLayer.CL_POINTER to GltfContainer or add MeshCollider
Need to optimize models for scene limits? See the optimize-scene skill for triangle budgets and LOD patterns. Need animations from your model? See the animations-tweens skill for playing GLTF animation clips with Animator.

Model Best Practices

  • Keep models under 50MB per file for good loading times
  • Use .glb format (binary GLTF) — smaller than .gltf
  • Optimize triangle count: aim for under 1,500 triangles per model for small props
  • Use texture atlases when possible to reduce draw calls
  • Models with embedded animations can be played with the Animator component
  • Test model orientation — Decentraland uses Y-up coordinate system
  • Materials in models should use PBR (physically-based rendering) for best results

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

36.12%
按下载量换算69

Claude

30.22%
按下载量换算57

Cursor

15.87%
按下载量换算30

Gemini CLI

9.57%
按下载量换算18

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

需要联网

该 Skill 可能需要联网访问来源站点、仓库或外部 API;具体网络访问范围需要结合源码和 README 复核。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills