3D printing

// Parametric box with friction-fit lid. Units: mm.
 
// inside length
inner_x = 60;   // [20:1:200]
// inside width
inner_y = 40;   // [20:1:200]
// inside height
inner_z = 25;   // [10:1:100]
// wall thickness, a multiple of line width
wall = 2;       // [1.2:0.2:4]
// floor thickness
floor_t = 2;    // [1:0.2:4]
// fit gap per side; tune with a test print
clear = 0.3;    // [0.1:0.05:0.6]
// height of the lid lip
lip_h = 6;      // [3:1:15]
// outer corner radius
r = 3;          // [0.5:0.5:8]
 
$fn = 48;
outer_x = inner_x + 2 * wall;
outer_y = inner_y + 2 * wall;
 
// rounded-rectangle slab, x by y, height h
module rounded(x, y, h, rad) {
  linear_extrude(h)
    offset(r = rad) offset(delta = -rad)
      square([x, y]);
}
 
module box() {
  difference() {
    rounded(outer_x, outer_y, inner_z + floor_t, r);
    translate([wall, wall, floor_t])
      rounded(inner_x, inner_y, inner_z + 1,
              max(r - wall, 0.01));
  }
}
 
// lid prints plate-down, lip pointing up: no supports
module lid() {
  lx = inner_x - 2 * clear;
  ly = inner_y - 2 * clear;
  rounded(outer_x, outer_y, wall, r);
  translate([wall + clear, wall + clear, wall])
    difference() {
      rounded(lx, ly, lip_h, max(r - wall - clear, 0.01));
      translate([wall, wall, -1])
        rounded(lx - 2 * wall, ly - 2 * wall, lip_h + 2, 0.01);
    }
}
 
box();
translate([0, outer_y + 10, 0]) lid();

<context>
I need a printed [part, e.g. replacement knob for a stove's D-shaft].
Printer: [printer: open FDM | enclosed FDM | MSLA resin | print service].
I'm new to CAD.
</context>
 
<task>
Interview me the way a designer scopes a functional printed part,
before any modeling. Don't design anything yet.
</task>
 
<requirements>
- At most 10 questions, grouped: function and loads; mating parts
  and measurements; environment (heat, sun, water, food contact);
  printer and material; hardware on hand.
- For each question: why it matters to the design, and typical
  answers.
- Tell me exactly what to measure with calipers and how (e.g. a
  shaft's diameter at three points, the D-flat across), to 0.05 mm.
</requirements>
 
<process>
Ask all the questions in one message, then stop and wait. When I
answer, restate the requirements as a numbered list with values,
write [NEEDS MEASUREMENT] for any dimension I haven't given, and
end with a list of the assumptions you made.
</process>

<materials>
[paste the requirements list from stage 1]
</materials>
 
<task>
Propose [number of approaches: 2 | 3] materially different design
approaches for this part, so I can choose one to model.
</task>
 
<requirements>
- The approaches differ in how the part is made or held together:
  one piece or two; printed threads, heat-set inserts or nut traps;
  print orientation.
- For each: print orientation and why (layer direction vs. the
  load); supports needed; material; hardware; fit and clearance per
  interface; rough print time; the main way it fails.
- Apply the DfAM rules: overhangs within 45° of vertical, bridges
  under about 10 mm, walls at least 2 lines wide.
- Never invent numbers, names, quotes or sources: write
  [NEEDS MEASUREMENT] for missing dimensions and [NEEDS SOURCE] for
  datasheet values.
</requirements>
 
<output_format>
A short section per approach, then a comparison table (strength,
printability, fit accuracy, print time, cost) and a recommendation.
</output_format>
 
<process>
Stop after the recommendation and wait for my choice. End with a
list of the assumptions you made.
</process>

<materials>
[paste the requirements from stage 1 and the chosen approach from
stage 2; add photos of the mating parts next to a ruler]
</materials>
 
<context>
Part: [part, e.g. knob for a 6.0 mm D-shaft with a 4.5 mm flat].
Printer: [printer, e.g. bed-slinger, 0.4 mm nozzle, 250 × 210 ×
220 mm build volume].
Material: [material: PLA | PETG | ASA | TPU | nylon].
Slicer: [slicer and version, e.g. PrusaSlicer 2.9].
I'm new to [CAD tool: OpenSCAD | CadQuery | build123d]; comment the
code so I can change it.
</context>
 
<task>
Write a parametric model of this part in that CAD tool, with its
print orientation and starting slicer settings, ready for a test
print.
</task>
 
<requirements>
- Every dimension a named parameter at the top, with a comment and
  units (mm); in OpenSCAD add Customizer ranges.
- One clearance parameter per interface, per side; default for
  [fit: press ≈ 0.15 mm | sliding ≈ 0.3 mm | loose ≈ 0.45 mm].
  A hole or slot grows by twice the clearance.
- Derived values computed from parameters, never typed in.
- Model it in print orientation, flat face on z = 0: no overhang
  beyond 45° from vertical; horizontal holes teardrop-shaped;
  bottom edges chamfered 0.5 mm (no bottom fillets); inside corners
  filleted.
- Walls at least [minimum wall, e.g. 2 mm], a multiple of the line
  width.
- Hardware sizes (insert holes, screw heads) from datasheets: write
  [NEEDS SOURCE] if you don't have the datasheet value.
- CadQuery or build123d: only documented API calls; say which
  version you assume; export STEP and STL.
- Never invent numbers, names, quotes or sources: write
  [NEEDS MEASUREMENT] for any dimension I didn't give.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every fit dimension is my measurement plus the clearance, applied
  on each side.
- In the chosen orientation the load runs along the layers, not
  across them.
- It prints without supports, or the draft says where they go.
- Changing one parameter (e.g. the shaft diameter) updates
  everything that depends on it.
- The code uses only real functions and would render cleanly.
</quality_bar>
 
<output_format>
The code in one block; a parameter table (name, value, what it
controls); the print orientation and why; a slicer settings table
(layer height, walls, top and bottom layers, infill, supports,
brim); the three features most likely to need tuning after a test
print.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change
your answer; otherwise go on. End with a list of the assumptions
you made.
</process>

<materials>
<brief>
[paste the requirements from stage 1]
</brief>
<design>
[paste the code and notes from stage 3, plus screenshots from three
angles in print orientation]
</design>
</materials>
 
<task>
Review this part for printability and fit before I print it, as a
DfAM review. Critique only: don't rewrite the code.
</task>
 
<requirements>
- Score each 1-5 with evidence from the code: dimensions vs.
  measurements; parametric (one change propagates); fits and
  clearances; orientation vs. loads; overhangs and supports; walls
  and minimum features; hardware sizes; code correctness (real
  functions, renders cleanly).
- Recompute every derived dimension yourself, including the
  clearance on both sides of each hole and slot.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where a datasheet value is assumed.
</requirements>
 
<quality_bar>
Before answering, check your review against:
- Every score cites a line of code or a number.
- Each fix is concrete: a parameter value, a geometry change or a
  slicer setting.
- It names the one feature to test-print first, as a thin slice.
</quality_bar>
 
<output_format>
A rubric table; fixes ranked by the risk of a failed or unusable
print; a 10-minute test print of the riskiest feature.
</output_format>
 
<process>
End with a list of the assumptions you made.
</process>

Compare [number: 2 | 3 | 4] materials for [part and conditions, e.g. a
bracket in a parked car, summer sun, light load] on heat, UV,
creep, ease of printing and what my printer needs (enclosure,
hardened nozzle, drying). Recommend one, then stop. Give typical
ranges only and say when the spool label or datasheet wins.

<materials>
[paste the spool labels or datasheets of filaments you already own]
</materials>
 
<context>
Part: [part and use, e.g. a hook for a garden hose, 3 kg, constant].
Where: [conditions: indoors | hot car | outdoor sun | wet | near food].
Printer: [printer, e.g. open-frame bed-slinger, brass 0.4 mm nozzle,
no enclosure].
</context>
 
<task>
Compare candidate materials for this part and recommend one, so I
buy the right filament before designing.
</task>
 
<requirements>
- Compare 2-4 of: PLA, PETG, ABS, ASA, TPU, nylon, PC,
  fiber-filled, standard or tough resin.
- For each: heat resistance; UV and water; creep under a constant
  load; layer adhesion; ease of printing; printer needs (enclosure,
  all-metal hotend, hardened nozzle, drying).
- Temperatures as typical ranges only; the label or datasheet wins.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] for any property you can't support.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The pick survives the worst condition I named (heat, sun, load).
- It prints on my printer as described, or says what to change.
- Food, child or safety use is flagged as needing a real test, not
  settled by the material choice.
</quality_bar>
 
<output_format>
Comparison table (material, heat, UV and water, creep, ease,
printer needs), then the recommendation in two sentences.
</output_format>
 
<process>
Stop after the recommendation and wait. End with a list of the
assumptions you made.
</process>

Write an OpenSCAD tolerance test: five 10 mm pins, and a bar with five
holes for per-side clearances of 0.1 to 0.5 mm (10.2 to 11.0 mm),
each value engraved, flat on the bed, no supports. Check that every
hole = pin + 2 × clearance, and tell me how to read the result as
press, sliding and loose fits.

<context>
Printer: [printer, e.g. bed-slinger, 0.4 mm nozzle].
Material: [material: PLA | PETG | ASA | TPU | nylon].
Slicer: [slicer and version, e.g. OrcaSlicer 2.3].
I'll use the result to set the clearance parameter in every design.
</context>
 
<task>
Write an OpenSCAD tolerance test coupon that shows which clearance
gives a press, sliding and loose fit on this printer.
</task>
 
<requirements>
- Five 10 mm pins, 8 mm tall, on a bar; a second bar 80 × 16 × 5 mm
  with five through holes for per-side clearances of
  [clearance steps, e.g. 0.1, 0.2, 0.3, 0.4, 0.5 mm], at the pins'
  spacing.
- Engrave each clearance value beside its hole, 0.6 mm deep.
- Flat on the bed, no supports, bottom edges chamfered 0.4 mm.
- Pin diameter and the clearance list as parameters.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Each hole diameter = pin + 2 × clearance (10.2 to 11.0 mm with
  the default steps).
- Both bars print in under about 30 minutes at normal settings.
- The engraved text is large enough to print (about 5 mm tall).
</quality_bar>
 
<output_format>
The code; then how to read the result, and how to measure the pins
with calipers to separate printer error from design error.
</output_format>
 
<process>
End with a list of the assumptions you made.
</process>

Give starting [slicer and version, e.g. PrusaSlicer 2.9] settings for
[part and use, e.g. a wall bracket holding 4 kg] in [material, e.g.
PETG], 0.4 mm nozzle: layer height, walls, top and bottom layers,
infill, supports, brim, seam, plus modifiers around the screw holes.
Use my slicer's own setting names and mark any you're unsure of.

<materials>
[paste or attach screenshots of the part in the slicer, or the code
and dimensions]
</materials>
 
<context>
Part: [part and use, e.g. a wall bracket holding 4 kg].
Material: [material, e.g. PETG]. Printer: [printer, e.g. direct-drive
bed-slinger, 0.4 mm nozzle]. Slicer: [slicer and version, e.g.
OrcaSlicer 2.3].
Priority: [priority: strength | surface finish | speed | accuracy].
</context>
 
<task>
Write a starting slicer profile for this part so the first print is
usable.
</task>
 
<requirements>
- Cover layer height, line width, walls, top and bottom layers,
  infill density and pattern, supports (type and where), brim, seam
  position and cooling; speeds only where they matter for this part.
- More walls before more infill for strength; modifiers for local
  changes (around holes and bosses, the cosmetic face).
- Use my slicer's own name for each setting; write [VERIFY] where
  you're unsure it exists in my version.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The settings serve my priority, and the trade-off is named.
- Layer height is at most about 75% of the nozzle diameter; walls
  are a whole number of lines.
- The cosmetic face has no supports and no seam.
</quality_bar>
 
<output_format>
Table: setting (my slicer's name), value, why. Then the modifiers.
Then what to inspect on the first print.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change
your answer; otherwise go on. End with a list of the assumptions
you made.
</process>

I switched to [material, e.g. a new brand of PETG] on [printer, e.g. a
direct-drive bed-slinger] with [slicer and version, e.g. OrcaSlicer
2.3]. Give the calibration sequence in order (temperature, flow,
pressure advance, retraction, max volumetric speed, cooling), one
variable per test, and say where each result goes in my slicer.

<context>
Material: [material, e.g. a new brand of PETG, opened today].
Printer: [printer and extruder, e.g. bed-slinger, direct drive, 0.4
mm brass nozzle]. Slicer: [slicer and version, e.g. OrcaSlicer 2.3].
Symptoms so far: [symptoms, e.g. stringing and rough overhangs].
</context>
 
<task>
Write a calibration sequence for this filament, in order, that ends
with a filament profile I can trust.
</task>
 
<requirements>
- Order: dry it first if it absorbs moisture; then temperature
  tower; flow (extrusion multiplier); pressure or linear advance;
  retraction; max volumetric speed; cooling.
- For each step: the test print (my slicer's built-in calibration
  if it has one), what to look for, how to read it, where to save
  the value.
- Use my slicer's own names; write [VERIFY] where you're unsure a
  menu or setting exists in my version.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- One variable per test, and each result is saved before the next.
- Temperatures start from the spool label's range.
- Firmware differences (pressure advance vs. linear advance) are
  called out.
</quality_bar>
 
<output_format>
Numbered steps: test, settings, what good looks like, where to save
it. Then a one-line-per-step checklist to print and tick.
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change
your answer; otherwise go on. End with a list of the assumptions
you made.
</process>

I generated a figurine and exported [format: OBJ | GLB | STL]. Target:
[height, e.g. 75 mm] on [process: FDM 0.4 mm | MSLA resin]. Give the
Blender cleanup step by step: units and scale, non-manifold repair,
thin parts at final scale, a flat base, hollowing and drain holes for
resin, 3MF export. Name the menus; mark any you're unsure of.

<materials>
[paste or attach screenshots of the mesh from the front, side and
below, with its polygon count]
</materials>
 
<context>
Source: [tool and format, e.g. an image-to-3D generator, OBJ with
textures]. Target: [height, e.g. 75 mm tall] on
[process: FDM 0.4 mm | MSLA resin | SLS print service].
Blender: [Blender version, e.g. 4.5] with the 3D-Print Toolbox.
</context>
 
<task>
Write a step-by-step Blender plan that turns this mesh into a
printable file.
</task>
 
<requirements>
- Steps: units in mm and scale to the target height; fix
  non-manifold edges, flipped normals, floaters and
  self-intersections; find parts thinner than
  [minimum thickness, e.g. 1.5 mm for FDM, 0.8 mm for resin] and
  thicken or remove them; add a flat base; for resin, hollow to
  about 2-2.5 mm walls with drain holes near the base; split with
  alignment pins if it won't fit or needs fewer supports; export
  3MF.
- Name the menus and operators; write [VERIFY] where you're unsure
  they exist in my Blender version.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The plan ends with a manifold check that reports zero problems.
- Thickness is checked at the final printed scale, not on import.
- Resin parts have drain holes and no closed cavities.
</quality_bar>
 
<output_format>
Numbered steps, each with its operator and what to check before
moving on.
</output_format>
 
<process>
End with a list of the assumptions you made.
</process>

My model is [model size, e.g. 320 × 180 × 260 mm]; my build volume is
[build volume, e.g. 250 × 210 × 220 mm]. Plan the split for
[process: FDM | resin]: cut planes on flat, hidden faces, alignment
(pins, keys or dovetails), glue gap, orientation per part. Check that
every part fits the build volume, brim included.

<materials>
[paste or attach screenshots of the model from three sides with its
overall dimensions]
</materials>
 
<context>
Model: [model, e.g. a 320 mm tall bust, decorative]. Build volume:
[build volume, e.g. 250 × 210 × 220 mm]. Process:
[process: FDM | MSLA resin]. Tools: [tools, e.g. Blender 4.5 and
PrusaSlicer 2.9].
</context>
 
<task>
Plan how to split this model into printable parts that align and
glue cleanly.
</task>
 
<requirements>
- Cut on flat, hidden faces, with as few parts as fit the build
  volume and the fewest supports per part.
- Alignment: [alignment: 4 mm pins | dowels | keyed joints | dovetails],
  with hole sizes using the clearance from my tolerance test.
- Glue gap 0.1-0.2 mm; print orientation for each part.
- Give the cut steps in Blender or the slicer's cut tool; write
  [VERIFY] for any menu you're unsure of in my version.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- Every part, with its brim, fits the build volume as oriented.
- No seam crosses a face that has to look clean.
- Pin holes allow for clearance on both sides.
</quality_bar>
 
<output_format>
Table: part, size as oriented, orientation, supports, alignment.
Then the cut steps.
</output_format>
 
<process>
End with a list of the assumptions you made.
</process>

[paste or attach 2-4 sharp photos with a ruler: whole part, defect,
first layer] Printer [printer, e.g. bed-slinger, Bowden], slicer
[slicer and version, e.g. Cura 5.8], material [material, e.g. PETG],
settings [settings, e.g. 240/80 °C, 0.2 mm, 60 mm/s]. Name the defect,
rank causes, give one change per test print; never change two at once.

<materials>
[paste or attach 2-4 sharp, well-lit photos: the whole part, the
defect up close, the first layer, with a ruler or coin for scale]
</materials>
 
<context>
Printer: [printer and extruder, e.g. bed-slinger, Bowden extruder].
Slicer: [slicer and version, e.g. UltiMaker Cura 5.8].
Material: [material and storage, e.g. PETG, opened 3 months ago,
stored in the open]. Settings: [settings, e.g. nozzle 240 °C, bed
80 °C, 0.2 mm layers, 60 mm/s, retraction 5 mm, fan 50%].
Changed since the last good print: [changes, e.g. new spool, the
printer moved next to a window].
</context>
 
<task>
Diagnose this print from the photos and give me a test plan, so I
fix it in as few prints as possible.
</task>
 
<requirements>
- Name the defect by its standard term (stringing, warping,
  elephant's foot, under-extrusion, layer shift, ringing, pillowing,
  layer splitting).
- Rank the likely causes, saying which evidence in the photos or
  settings points to each.
- One change per test print, in order, with the result to expect.
- Never invent numbers, names, quotes or sources: write
  [NEEDS SOURCE] where support is missing, and [VERIFY] for any
  setting name you're unsure of in my slicer version.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The diagnosis matches what the photos show, not just the most
  common defect.
- Cheap, reversible fixes (drying, cleaning the bed, temperature)
  come before hardware changes.
- No test changes more than one setting.
</quality_bar>
 
<output_format>
Defect name; causes ranked with evidence; a numbered test plan
(change, why, expected result).
</output_format>
 
<process>
Ask me up to 3 questions first if anything missing would change
your answer; otherwise go on. End with a list of the assumptions
you made.
</process>

Make me a phone stand I can 3D print.

<context>
Printer: bed-slinger FDM, 0.4 mm nozzle, 250 × 210 × 220 mm build
volume. Material: PETG. Slicer: PrusaSlicer 2.9. First time with
OpenSCAD.
Phone in its case, measured with calipers: 78.2 mm wide, 11.8 mm
thick. It stands in portrait on a desk while charging over USB-C.
</context>
 
<task>
Write a parametric OpenSCAD phone stand that I can print and adjust
after one test print.
</task>
 
<requirements>
- Viewing angle 65° from horizontal. Lip 10 mm tall with a 14 × 8 mm
  cable slot. Base 90 mm wide × 100 mm deep so it won't tip; two
  10 mm recesses, 1 mm deep, for rubber feet.
- Parameters at the top with Customizer ranges: phone thickness,
  clearance per side (default 0.5 mm, because the phone drops in),
  angle, lip height, cable slot size, wall (default 3 mm).
- Print on its side so the layer lines run along the bending load,
  with no supports; chamfer bed edges 0.5 mm; fillet the inside
  corner where the back meets the base (r = 5 mm).
- Never invent numbers, names, quotes or sources: write
  [NEEDS MEASUREMENT] for any dimension I haven't given.
</requirements>
 
<quality_bar>
Before answering, check your draft against:
- The phone slot = phone thickness + 2 × clearance.
- The phone's center of mass sits over the base, so it won't tip.
- No overhang beyond 45° in the print orientation.
- Every derived dimension is computed from parameters.
</quality_bar>
 
<output_format>
The code, then a parameter table, why that orientation, and
starting PrusaSlicer settings for PETG.
</output_format>
 
<process>
End with a list of the assumptions you made.
</process>

// Phone stand. Units: mm. Prints on its side.
phone_t = 11.8;  // [8:0.1:16] phone thickness, in its case
clear   = 0.5;   // [0.2:0.1:1.5] clearance per side
angle   = 65;    // [50:1:80] viewing angle from horizontal
lip_h   = 10;    // [5:1:20] lip height
wall    = 3;     // [2:0.5:5] wall thickness
width   = 90;    // [70:1:120] stand width
slot_w  = phone_t + clear;   // phone slot
...
module profile() { ... }     // side profile, in the XY plane
linear_extrude(width) profile();   // profile flat on the bed
 
Orientation: on its side, so each layer is the whole profile
and bending runs along the lines, not across them.
Tip check: the phone's center of mass is about 34 mm behind
the lip, well inside the 100 mm base.
PrusaSlicer 2.9, PETG: 0.2 mm layers, 4 perimeters, 5 top and
bottom layers, 20% gyroid, no supports, no brim.
Tune after a test print: clearance, lip height, angle.
Assumptions: the case has no ring grip; the desk is flat; ...

slot_w adds the clearance once, but clear is per side. Change it to
phone_t + 2 * clear and check every other fit in the code for the
same mistake. Then add a test_slice parameter that cuts the stand to
a 10 mm wide slice, so I can test the slot in a 15-minute print
before printing the whole stand.

Follow-up critique prompts
 
- "List every hard-coded number in this code and turn the ones
  that depend on other dimensions into derived values."
- "Which feature fails first under [load, e.g. 2 kg hanging from
  the hook]? Change the orientation or geometry to fix it."
- "Where would a slicer add supports? Redesign to remove them."
- "Make a 10-minute test print of just the
  [feature: hinge | thread | fit]."
- "Check this against the OpenSCAD documentation: is every
  function and argument real?"
- "The lid is too tight by about 0.2 mm. Which parameters change,
  and to what?"