Claude Cursor Skill

assemble-scene

Use when composing PlayCanvas entity hierarchies from models, colliders, rigid bodies, and scripts, so each object has one semantic root and clear lifecycle.

LLM Mart · 0 points · 1 views 0 listing impressions 0 install-command copies
Download playcanvas-skills-skills_assemble-scene-8fd2b26.zip · 2 KB

Install

skills CLI npx skills add https://github.com/playcanvas/skills/tree/main/skills/assemble-scene
Claude Code claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install playcanvas-skills@llmmart
Git git clone https://github.com/playcanvas/skills.git

The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole playcanvas/skills collection as a plugin from our marketplace. Git is the plain clone.

Skill manifest

Scene assembly

Each scene object has one named root. Put position, heading, rigidbody, collision, and scripts on the root, and the calibrated visual from calibrate-model below it. Apply calibration once; if the tuning record is missing, calibrate first.

  • Name roots by role, with predictable names for repeated instances.
  • Size colliders from measured bounds and keep them aligned with render bounds.
  • Space roots by the scaled footprint in the tuning record, not by pivot position. A non-90° yaw rotates the footprint, so reserve its larger horizontal extent.
  • Keep UI, effects, cameras, and lights outside model hierarchies.
  • Remove the root through its surface's lifecycle and unregister events it owns.
  • Rigid bodies need Ammo loaded first (build-app).

Check placement

After a rendered frame, measure each support or attachment at the object's actual position, using a mount point, raycast, or geometry sample read through inspect-runtime. A positive gap floats; a negative gap penetrates. Overlapping world AABBs between unrelated roots are a failure unless two different views prove the overlap is empty space.

Read only the reference matching the code: direct Engine, React, or Web Components.

Files (skills)
  • references
    • direct-engine.md 439 B
      Loading…
    • react.md 1023 B
      Loading…
    • web-components.md 892 B
      Loading…
  • SKILL.md 1.5 KB
    ---
    name: assemble-scene
    description: Use when composing PlayCanvas entity hierarchies from models, colliders, rigid bodies, and scripts, so each object has one semantic root and clear lifecycle.
    ---
    
    # Scene assembly
    
    Each scene object has one named root. Put position, heading, `rigidbody`, `collision`, and
    scripts on the root, and the calibrated visual from `calibrate-model` below it. Apply calibration
    once; if the tuning record is missing, calibrate first.
    
    - Name roots by role, with predictable names for repeated instances.
    - Size colliders from measured bounds and keep them aligned with render bounds.
    - Space roots by the scaled footprint in the tuning record, not by pivot position. A non-90° yaw
      rotates the footprint, so reserve its larger horizontal extent.
    - Keep UI, effects, cameras, and lights outside model hierarchies.
    - Remove the root through its surface's lifecycle and unregister events it owns.
    - Rigid bodies need Ammo loaded first (`build-app`).
    
    ## Check placement
    
    After a rendered frame, measure each support or attachment at the object's actual position, using
    a mount point, raycast, or geometry sample read through `inspect-runtime`. A positive gap floats; a negative gap penetrates.
    Overlapping world AABBs between unrelated roots are a failure unless two different views prove the
    overlap is empty space.
    
    Read only the reference matching the code: [direct Engine](references/direct-engine.md),
    [React](references/react.md), or [Web Components](references/web-components.md).
    

Comments (0)

Sign in to join the conversation.

No comments yet.

Reviews (0)

No reviews yet.