[aldoruizluna]

What

Phase 8b, lane P8-TWINS-A (Digital Twins MES programme). xy-joint/xy-joint.graph.json is a node-graph twin of xy-joint/main.py, in graph format 1.1.

• Declaration. It is declared as graph_file on the xy_joint mode, next to the script.
• Which one renders. Owner decision D5 applies: graph twins coexist with scripts and are parity-checked. main.py stays the rendered source and the oracle.
• Unchanged: frames, presets and parameters. docs/README.md gains a "Graph twin" section (en) and a paragraph in the Resumen (es).
• How the twin is built:
  • It has the same parameters, declared in parameters.
  • It has the same derivations:
    • the idler_width and d_end clamps are ternary pairs;
    • the belt levels, the raised element's boss, the post's top (max) and its inboard edge (min) are ternaries.
  • It performs the same operations in the same order.
  • The script's if mirrored: seat ring is a select.
  • The right-hand joint is body.mirror("YZ"), expressed as reflect YZ chosen by select on mirrored.
• Seam. D's back-slot screw runs along +Y. A Z cylinder laid down with rotate x −90 and spun with rotate y −90 puts the B-rep seam where makeCylinder puts it (x direction +Z). An XZ profile_circle + extrude would put it at +X, which measured up to 0.0019 mm off on some diameters.

Evidence

Run locally on keystone main eeb46798, with OpenSCAD 2026.02.13:

y4d-spec check ./xy-joint --render --require-openscad --parity --openscad-path libs --openscad-path . -v
parity (…, cadquery vs graph): ok — Meshes are identical within 0.000000mm tolerance.   (defaults, left_350, right_350 [mirrored], left_10mm_idler, right_10mm_idler [mirrored])
y4d-spec check: cartridges=1 failures=0 notes=1 … renders=10 presets=8 … parity=5/5 ok, warn=0, exempt=0, placement=0, failures=0 frames=50/50 ok, unverified=0, failures=0

The one note is pre-existing: the frame grammar uses ASM-1 v1.1.0 let.

• Structural check: rc 0.
• pytest .github/scripts: 59 passed.
• assembly_check.py + y4d-spec assembly check --collision: checked=2 failed=0. Assembly A has 235/235 placed, 292/292 mates, 23/23 poses and errors=0.

Assembly A's digest moves. It goes from cb80b6790dbc7047… (solid main 345bc45) to 85362cde6e9f62132f247de968efe6f75c385eb9ecfac3ea41ee2e8d37a6ac3c (this PR alone). The keystone fixture follows later, in one refresh the coordinator schedules.

Landing

SPEC_PIN 142db18 already carries graph format 1.1. Draft: the coordinator merges.

🤖 Generated with Claude Code