Box Styles / ECMA A20.20.01 testing

# Reference ingestion — fEFCO_0211_reverse_friction_tuck_box_50×50×110_mm
Source class: **pacdora** (die_maker outranks pacdora when rules are tuned)

## Reference measurements
```json
{
 "blank_w_mm": 212.0,
 "blank_h_mm": 229.6,
 "panel_scores_x": [
  13.0,
  63.0,
  113.0,
  163.0
 ],
 "panel_widths": [
  50.0,
  50.0,
  50.0
 ],
 "body_scores_y": [
  60.0,
  169.0
 ],
 "cut_arc_radii": {},
 "bleed_offset_mm": 5.0,
 "dim_table": {
  "inside": "49(L) \u00d7 49(W) \u00d7 107.8(H) mm",
  "outside": "51.4(L) \u00d7 51.4(W) \u00d7 112.6(H) mm",
  "manufacture": "50(L) \u00d7 50(W) \u00d7 110(H) mm",
  "material": "F-flute",
  "thickness": "1.2 mm",
  "design_area": "212 \u00d7 229.6 mm"
 },
 "notes": [
  "PDF; 125 colour-classified segs (clipped to blank); text keys: ['inside', 'outside', 'manufacture', 'material', 'thickness', 'design_area']"
 ]
}
```

## Comparison vs voom-dieline
| metric | reference | ours | delta |
|---|---|---|---|
| blank_w_mm | 212 | 213.75 | +1.75 |
| blank_h_mm | 229.6 | 242 | +12.40 |
| panel_widths | [50.0, 50.0, 50.0] | [50.0, 50.0, 50.0] | — |
| body_scores_y | [60.0, 169.0] | [66.0, 176.0] | DIFFERS |
| cut_arc_radii | {} | {5.0: 4, 3.0: 8, 2.0: 4} | DIFFERS |
| bleed_offset_mm | 5 | 3 | -2.00 |

### Rules our generator applied
- panel_widths: shrink_last: [50.0, 50.0, 50.0, 48.75] — glue-seam panel −(5/6)t pulls the joint square (Pacdora-calibrated)
- lid_taper: lid sides taper 1.10/side — lid rotates past the side panels
- tuck_inset+shoulder: tuck inset 5.85/side = taper + sliding clearance 1.75 + shoulder land 3.00, undercut 0.5
- dust_gap_to_lid: dust edge facing lid cleared 3.75 — folded lid wall passes it
- dust_relief: 2.25 relief nick at dust base — 3 board layers meet at that corner
- tangent_tip_arc: dust tip radii tangent to the angled sides (no rule kinks)
- crease_vs_knife: creases drawn only where hinges exist; interrupted at notches/slits
- slit_length: lock slits 6.75 deep — shoulder land + sliding clearance

## LLM analysis (DRAFT — gated by human review + test cut)
### 1. Style Identification (FEFCO/ECMA Code Family)

Based on the measurements, this dieline appears to be a **0201** style box, which is a standard regular slotted container (RSC) with a lid and tuck-in flap. The dimensions and structure are consistent with this style.

### 2. Engineering Differences that Matter

#### Reference vs. Generator
- **Blank Height and Width:**
  - **Reference:** `blank_h_mm = 229.6`, `blank_w_mm = 212`
  - **Generator:** `blank_h_mm = 242.0`, `blank_w_mm = 213.75`
  - **Difference:** The generator's blank height is significantly larger (12.4mm), which could affect the overall fit and appearance of the box.

- **Panel Scores:**
  - **Reference:** `panel_scores_x = [13, 63, 113, 163]`
  - **Generator:** `panel_scores_x = [15.0, 65.0, 115.0, 165.0]`
  - **Difference:** The generator's scores are slightly shifted (2mm), which could affect the alignment of panels.

- **Body Scores:**
  - **Reference:** `body_scores_y = [60, 169]`
  - **Generator:** `body_scores_y = [66.0, 176.0]`
  - **Difference:** The generator's scores are also slightly shifted (6mm), which could affect the fit of the lid and body.

- **Bleed Offset:**
  - **Reference:** `bleed_offset_mm = 5`
  - **Generator:** `bleed_offset_mm = 3.0`
  - **Difference:** The generator's bleed offset is smaller, which could result in insufficient bleed for printing.

### 3. Concrete Rule-Change Hypotheses

#### Blank Height and Width
- **Hypothesis:** Adjust the blank height to match the reference.
  - **Proposal:** Set `blank_h_mm = 229.6` in the generator.
  - **Confidence:** High
  - **Test:** Measure the assembled box's height and ensure it matches the reference.

#### Panel Scores
- **Hypothesis:** Align panel scores with the reference.
  - **Proposal:** Set `panel_scores_x = [13, 63, 113, 163]` in the generator.
  - **Confidence:** High
  - **Test:** Assemble the box and check for proper alignment of panels.

#### Body Scores
- **Hypothesis:** Align body scores with the reference.
  - **Proposal:** Set `body_scores_y = [60, 169]` in the generator.
  - **Confidence:** High
  - **Test:** Assemble the box and check for proper fit of the lid and body.

#### Bleed Offset
- **Hypothesis:** Increase the bleed offset to match the reference.
  - **Proposal:** Set `bleed_offset_mm = 5` in the generator.
  - **Confidence:** High
  - **Test:** Print a test run and ensure there is sufficient bleed for trimming.

### 4. Questions to Ask the Die Maker

1. **Blank Height:**
   - Why is the reference blank height 229.6mm, while our generator produces 242.0mm? Is this intentional, or should we adjust the generator?

2. **Panel Scores:**
   - Are the panel scores at [13, 63, 113, 163] critical for proper alignment, or can they be adjusted to [15.0, 65.0, 115.0, 165.0]?

3. **Body Scores:**
   - Are the body scores at [60, 169] necessary for the correct fit of the lid and body, or can they be adjusted to [66.0, 176.0]?

4. **Bleed Offset:**
   - Is a bleed offset of 5mm sufficient for printing, or should we use the generator's 3.0mm? What are the implications of using a smaller bleed offset?

5. **General Fit and Functionality:**
   - Are there any specific tolerances or adjustments needed to ensure the box fits and functions correctly when assembled?

_model: qwen2.5:72b_