Claude-skill-registry godot-extract-to-scenes

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/godot-extract-to-scenes" ~/.claude/skills/majiayu000-claude-skill-registry-godot-extract-to-scenes && rm -rf "$T"
manifest: skills/data/godot-extract-to-scenes/SKILL.md
source content

Extract Code-Created Objects to Scenes

Core Principle

Every node should be a scene. Code-created objects with

.new()
should be rare exceptions for truly dynamic content.

What This Skill Does

Finds patterns like:

var timer = Timer.new()
timer.wait_time = 2.0
timer.one_shot = true
add_child(timer)

Transforms to:

@onready var timer: Timer = $Timer
# Timer is now a scene node, configured in editor

Detection Patterns

Scans for:

  • Timer.new()
  • Area2D.new()
    ,
    CollisionShape2D.new()
  • Sprite2D.new()
    ,
    AnimatedSprite2D.new()
  • AudioStreamPlayer.new()
    ,
    AudioStreamPlayer2D.new()
  • Control.new()
    ,
    Label.new()
    ,
    Button.new()
  • Any
    Node*.new()
    pattern

When to Use

You're Building New Features

Your code creates objects dynamically that should actually be scenes.

You're Refactoring Legacy Code

Old code has

.new()
calls that make nodes invisible in editor.

You Want Inspector Visibility

You want to configure properties in the Godot editor, not in code.

Process

  1. Scan - Find all
    .new()
    patterns in .gd files
  2. Analyze - Determine node type, properties, parent relationships
  3. Generate - Create .tscn scene files with proper configuration
  4. Update - Replace code creation with @onready references
  5. Commit - Git commit per component extracted

Example Transformation

Before:

# player.gd
func _ready():
    var hitbox = Area2D.new()
    hitbox.name = "Hitbox"
    var collision = CollisionShape2D.new()
    var shape = RectangleShape2D.new()
    shape.size = Vector2(32, 48)
    collision.shape = shape
    hitbox.add_child(collision)
    add_child(hitbox)

After:

# player.gd
@onready var hitbox: Area2D = $Hitbox
# Hitbox is now configured as scene node in player.tscn

Generated File:

# components/hitbox_component.tscn
[gd_scene load_steps=2 format=3]

[sub_resource type="RectangleShape2D" id="1"]
size = Vector2(32, 48)

[node name="Hitbox" type="Area2D"]

[node name="CollisionShape2D" type="CollisionShape2D" parent="."]
shape = SubResource("1")

What Gets Created

  • Component library in
    components/
    directory
  • .tscn scene files with proper node hierarchy
  • Preset .tres files for reusable configurations
  • Updated parent scripts with @onready references
  • Git commits documenting each extraction

Smart Detection

Skips intentional dynamic creation:

  • Objects created in loops (spawning enemies)
  • Objects created conditionally (different weapon types)
  • Objects with variable types (plugin systems)

Focuses on static objects:

  • Same object created every time
  • Properties set to constant values
  • Should exist in editor for visibility

Integration

Works with:

  • godot-split-scripts - Extract, then split large scripts
  • godot-add-signals - Use signals with extracted components
  • godot-refactor (orchestrator) - Runs as part of full refactoring

Safety

  • Every operation creates a git commit
  • Validation tests run after extraction
  • Auto-rollback on test failure
  • Original code preserved in git history

When NOT to Use

Don't extract if:

  • Object is created dynamically (in loop, based on data)
  • Object type varies at runtime
  • Object is truly temporary (created and destroyed frequently)
  • Code is already using PackedScene.instantiate()

These are legitimate uses of

.new()
and should remain as code.