Hardware Pipeline
Public Made by Adomby adom
The Fusion CAD to Component-page workflow: export a machine from Fusion via the adom-desktop bridge, convert STEP to a web-grade draco GLB with step2glb, and publish a wiki Component page with the live 3D viewer, using the current viewer, hero, and size contracts.
Hardware Pipeline
The end-to-end workflow for putting hardware on the wiki: a machine or workcell goes from a Fusion cloud design to a live Component page with the interactive 3D viewer. Three stages: Fusion export through the adom-desktop bridge, STEP to draco-compressed GLB through step2glb, and an adom-wiki Component page publish that follows the current viewer, hero, and size contracts.
This is the hardware sibling of molecule-pipeline. Molecules keep their pipeline (MP markers, machine pins, footprint/symbol invariants); machines need none of that, just clean geometry, a sane file layout, and the page contracts encoded here so nobody re-learns them from tribal knowledge.
Install it and ask your AI to "publish a workcell page"; the skill walks the three stages with the exact commands, the gotchas (async 202 conversion jobs, multipart pushes for big STEP files, the 8 MB / 4M vertex inline budget), and the lint-first habit that replaces memorizing any of it.
Reference pages built with this flow: adom/essemtec-puma, adom/kiosk-workcell.
# Hardware Pipeline
The end-to-end workflow for putting hardware on the wiki: a machine or
workcell goes from a Fusion cloud design to a live Component page with the
interactive 3D viewer. Three stages: Fusion export through the adom-desktop
bridge, STEP to draco-compressed GLB through step2glb, and an adom-wiki
Component page publish that follows the current viewer, hero, and size
contracts.
This is the hardware sibling of molecule-pipeline. Molecules keep their
pipeline (MP markers, machine pins, footprint/symbol invariants); machines
need none of that, just clean geometry, a sane file layout, and the page
contracts encoded here so nobody re-learns them from tribal knowledge.
Install it and ask your AI to "publish a workcell page"; the skill walks the
three stages with the exact commands, the gotchas (async 202 conversion jobs,
multipart pushes for big STEP files, the 8 MB / 4M vertex inline budget), and
the lint-first habit that replaces memorizing any of it.
Reference pages built with this flow: adom/essemtec-puma, adom/kiosk-workcell.