feat(0127-LegAnkleFootLandmarkSpecsCaptureCompare): leg/ankle/foot landmark capture, compare, and placement consume

[Ryan-AI-Studios]

What

Track 0127: frozen form-read gastroc/malleolus/arch/ball/Achilles ids → fuse XYZ (left Y overlay + back X/Z, no new DEPTH_PAIRS) → sidecar leg_foot_metrics.json → CLI blockout-leg-foot-compare + MCP 51 (mesh_proportion_blockout_leg_foot_compare) → new-id calf_cyl p0 Y after B6 + arch_soft Z only.

Why

Package A has knee/ankle + calf diameters + calf DEPTH, but not gastroc/malleoli/arch/ball/Achilles form-read. Agents were guessing calf belly Y and arch Z from 0108/0096 H-frac priors.

Honesty (N6)

LEG_FOOT_COMPARE_HONESTY = proportion_leg_foot_compare_not_mesh_or_print_success. Compare is authoring QA only — not mesh or print success. Capsule _delta_mm uses p0 (live calf_cyl center=null). Y/Z-only _suggest ignores delta.x. Hold 0108/0096/0071/0042/0056. No remake/D7 (Package A lacks new ids). Report stay 1.2.0. Recipe stay 1.4.0.

Tests

A–F suites + pin-retarget catalog 50→51 (61 asserts / 52 files). Four gates green locally (ledgerful verify).

---

Note

Medium Risk
Touches fuse XYZ and recipe calf/arch placement, so wrong consume could shift blockout geometry. Scope is authoring QA with hold-priors for most ids and no schema bump.

Overview
Adds track 0127 form-read landmarks (gastroc, malleoli, arch, Achilles, ball) through capture, fuse, authoring compare, and limited recipe consume.

New blockout-leg-foot-compare (CLI + MCP, catalog 51) writes leg_foot_compare.json and optional leg_foot_metrics.json. Compare is authoring QA only (LEG_FOOT_COMPARE_HONESTY); suggestions stay in a closed set, with soft_adjust only for gastroc Y and arch Z. Capsule deltas use p0; X is ignored.

Fuse overlays left-view Y and back-view X/Z for the new ids (no new DEPTH_PAIRS, no invented Y). Recipe consume is narrow: measured gastroc_med_* Y onto calf_cyl p0 after B6 (then split mid), and measured arch_apex_* Z onto arch_soft. Existing 0108/0096 priors and report/recipe schemas stay 1.2.0 / 1.4.0.

^{Reviewed by Cursor Bugbot for commit 8d80d1b. Bugbot is set up for automated code reviews on this repo. Configure here.}

[cursor]

Cursor Bugbot has reviewed your changes using default effort and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 8d80d1b. Configure here.}

[cursor]

Gastroc soft_adjust never clears

Medium Severity

_suggest treats any large delta.y or delta.z as soft_adjust for all soft-adjust ids, but consume is axis-specific (gastroc_med_* writes only Y onto capsule p0; arch_apex_* writes only Z). Capsule p0 sits at knee height while photo gastroc Z is belly height, so delta.z stays large after a successful Y consume and soft_adjust never settles to hold_priors.

Additional Locations (2)

• src/meshops/proportion/leg_foot_compare.py#L99-L109
• src/meshops/proportion/blockout_recipe.py#L3226-L3233

 

^{Reviewed by Cursor Bugbot for commit 8d80d1b. Configure here.}

[cursor]

Legacy calf capsule shortened

Low Severity

When no calf_taper_dist exists, _apply_measured_calf_cyl_y still runs _calf_split_mid and writes the result into calf_cyl.p1. That turns a full-shank one-cyl capsule into a proximal 42% stub and leaves a gap to the ankle. Sibling thigh-adduction logic correctly leaves legacy p1 at the ankle and only moves p0.

Additional Locations (1)

• src/meshops/proportion/blockout_recipe.py#L5541-L5559

 

^{Reviewed by Cursor Bugbot for commit 8d80d1b. Configure here.}