recipe-loom-subassets

Turn a working chunk of a graph — a script body, a kitbash, a repeated part — into a NAMED sub-asset with published parameters, then compose the model out of those instead of re-deriving shapes every

SKILL.md

GitHub

Loom subassets (recipe stub)

This is a pointer, not the recipe. The workflow lives in the engine and always matches the build you are connected to, so load it from the running editor before you act:

ngine.recipes {name: "authoring/loom-subassets"}

Follow the numbered steps it returns, and read its warnings first. Do not work from this stub alone.

Use it when

  • a world.script node has grown past the point where a human could tune it
  • the same part appears more than once in a model (a window, a wheel, a hull panel)
  • the node graph shows one opaque box instead of describing the asset
  • handing a model to someone who should not have to read the script that built it
  • detail has to sit ON a sloped or curved surface (shingles, panels, rivets)

Ops it names

loom.promote_script, graph.collapse_to_tool, graph.expand_tool, graph.edit_tool_def, loom.defs, loom.reconfigure, loom.nest, loom.place, loom.set_param, loom.author, worldgraph.describe_node, worldgraph.inspect_rows

Check each op's arguments with ngine.describe before the first call: the recipe is the order and the discipline, the live schema is the spelling.

Start the session with editor.context. If ngine.recipes cannot be reached the editor is not running or not connected: run /apeiron:status.