| name | bodyfit-ready-check |
| description | Run the body-fit gate on a fit spec (product dimensions for a specific body): completeness, anthropometric/ISO ranges, frame-PD optical alignment, print floor — eyewear first; dental/prosthetics/footwear share the pattern. Trigger: "do these glasses fit", "check this fit spec", "frame for a 63 PD". NOT a fitting: face scan, angles, and Rx compatibility are the optician's.
|
| kind | skill |
| license | MIT |
| runtimes | ["claude-code","codex","hermes"] |
bodyfit-ready-check
When to use
Any body-fit product spec claiming to fit a declared body — before geometry
is generated or printed.
What it does
- B1 completeness: every
data/bodyfit.yml dimension declared — undeclared
is not measured ⇒ fail.
- B2 ranges: each dimension inside its anthropometric/ISO-lineage range.
- B3 alignment: frame PD (lens_w + bridge) within the decentration budget of
the wearer's PD — the dispensing rule as a number.
- B4 print floor: declared min feature ≥ the product's snap threshold; the
geometry itself is then gated by
print-ready-check on the STL.
Example
python3 examples/eyewear/generate.py fitspec.json temple.stl
python3 pipeline/bodyfit_gate.py fitspec.json --json
python3 pipeline/ready_gate.py temple.stl --min-feature 3.0
Verification (eval-with-teeth)
Success may be declared ONLY when both gates exit 0. Held by tests that
mutate the golden spec five known-bad ways (missing dim, out-of-range PD,
misaligned frame, thin feature, unknown product) and assert each fails:
python3 -m pytest tests/test_m8_m11_m9.py -q
Safety
Table-fit, not fitted: population ranges say a spec is plausible, not that it
fits one specific face. The report says so.
Cross-runtime
Python + PyYAML; exit codes gate any CI.
