ScadPy
Programmatic CAD in Pure Python. — Documentation
ScadPy is a parametric modeling library for Python. Define 2D shapes and 3D solids with a fluent, chainable API: boolean operations, extrusions, fillets, patterns and topology queries. Export to STL, DXF, SVG or render interactively.
Built on Shapely for 2D geometry, trimesh for 3D meshes and NumPy for topology queries (vertex coordinates, edge normals, ring types and more). If you know OpenSCAD, the approach will feel familiar.
Installation
pip install scadpy
Requirements: Python ≥ 3.12.
Quick examples
# 2D — chamfered mounting plate
from scadpy import rectangle, circle, text
import numpy as np
PLATE_WIDTH = 80
PLATE_HEIGHT = 50
PLATE_THICKNESS = 10
HOLE_RADIUS = 4
HOLE_MARGIN = 10
CHAMFER_SIZE = 8
base = rectangle([PLATE_WIDTH, PLATE_HEIGHT])
plate = base.chamfer(CHAMFER_SIZE)
for position, normal in zip(base.vertex_coordinates, base.vertex_normals):
hole_center = position - HOLE_MARGIN * np.sqrt(2) * normal
plate -= circle(HOLE_RADIUS).translate(hole_center)
plate.to_screen()
# 3D — extruded mounting plate with label (continues from above)
TEXT_THICKNESS = 2
extruded_plate = plate.linear_extrude(PLATE_THICKNESS)
label = text("ScadPy", size=15).linear_extrude(TEXT_THICKNESS)
extruded_plate |= label.translate(z(PLATE_THICKNESS))
extruded_plate.to_screen()
# 3D — parametric ball bearing
from scadpy import circle, rectangle, sphere, x, y, GRAY, ORANGE
BALL_RADIUS = 3
RACE_RADIUS = 15
NB_BALLS = 11
CLEARANCE = 0.1
RING_HEIGHT = 7
RACE_THICKNESS = 10
groove = circle(BALL_RADIUS + CLEARANCE) | rectangle([BALL_RADIUS, RING_HEIGHT])
race = rectangle([RACE_THICKNESS, RING_HEIGHT]) - groove
bearing = race.radial_extrude(axis=y(), pivot=x(RACE_RADIUS)).color(GRAY)
ball = sphere(BALL_RADIUS).color(ORANGE)
bearing += ball.radial_pattern(count=NB_BALLS, axis=y(), pivot=x(RACE_RADIUS))
bearing.to_screen()
# 3D — dice
from scadpy import cuboid, sphere, x, y, z
SIZE = 20
dice = cuboid(SIZE)
pip = sphere(SIZE / 12).translate(z(SIZE / 2))
one = pip
two = pip.translate([SIZE / 4, SIZE / 4, 0]) + pip.translate([-SIZE / 4, -SIZE / 4, 0])
three = one + two
four = two + two.rotate(90, z())
five = one + four
six = four + pip.translate(x(SIZE / 4)) + pip.translate(x(-SIZE / 4))
dice -= (
one
+ two.rotate(90, x())
+ three.rotate(90, y())
+ four.rotate(-90, y())
+ five.rotate(-90, x())
+ six.rotate(-180, x())
)
dice.to_screen()
# 3D — storage box
from scadpy import square, x, z
SIZE_OUTER = 20
SIZE_INNER = 18
FILLET = 1
BASE_HEIGHT = 10
CUT_HEIGHT = 8
CAP_HEIGHT_OUTER = 1.5
CAP_HEIGHT_INNER = 3
CAP_OFFSET_X = 25
CUT_OFFSET_Z = 2
outer_base = square(SIZE_OUTER).fillet(FILLET).linear_extrude(BASE_HEIGHT)
inner_cut = square(SIZE_INNER).linear_extrude(CUT_HEIGHT).translate(z(CUT_OFFSET_Z))
base = outer_base - inner_cut
cap_outer = square(SIZE_OUTER).fillet(FILLET).linear_extrude(CAP_HEIGHT_OUTER)
cap_inner = square(SIZE_INNER).linear_extrude(CAP_HEIGHT_INNER)
cap = (cap_outer | cap_inner).translate(x(CAP_OFFSET_X))
storage_box = base + cap
storage_box.to_screen()
AI integration
ScadPy ships a machine-readable skill file (ai-skills.txt) that lets AI
assistants understand the full API without reading source code — signatures,
descriptions, parameters, return types, and usage examples, extracted directly
from the source.
Cheat sheet
Parameters shown in # comments are optional, with their default values.
2D — Shape
from scadpy import *
# primitives
circle(radius=3) # segment_count=64
polygon(points=[(-2, -2), (2, -2), (0, 2)])
rectangle(size=[6, 3])
Shape.from_dxf("file.dxf")
Shape.from_svg("file.svg")
square(size=4)
# boolean operations
s = square(size=4); c = circle(radius=3)
s | c # union
s - c # difference
s & c # intersection
s ^ c # symmetric difference
s + c # concat (no merge)
# transforms
s.chamfer(size=0.8) # vertex_filter=None, epsilon=1e-8
s.color(color=RED)
s.convexify() # part_filter=None
s.fill() # part_filter=None
s.fillet(size=0.8) # vertex_filter=None, segment_count=32, epsilon=1e-8
s.grow(distance=0.5) # part_filter=None
s.linear_cut(axis=x()) # pivot=0
s.linear_pattern(counts=4, steps=x(3)) # counts=[nx, ny], steps=[sx, sy]
s.linear_slice(thickness=2, direction=x()) # pivot=0, part_filter=None
s.mirror(normal=[1, 0]) # pivot=0
s.pull(distance=1.0) # pivot=0, vertex_filter=None
s.push(distance=1.0) # pivot=0, vertex_filter=None
s.radial_pattern(count=6) # angle=360, pivot=0
s.radial_slice(start=0, end=180) # pivot=0, part_filter=None
s.resize(size=[6, None]) # auto=False, pivot=None, vertex_filter=None
s.rotate(angle=30) # pivot=0, vertex_filter=None
s.scale(scale=[2, 0.5]) # pivot=0, vertex_filter=None
s.shrink(distance=0.5) # part_filter=None
s.translate(translation=[2, 1]) # vertex_filter=None
# features
s.bounds # [min_x, min_y, max_x, max_y]
s.bounding_box # → Shape (rectangle)
s.centroid # [cx, cy] — geometric centroid
s.is_empty # bool
# topology — coordinates & attributes
s.are_vertices_convex # (n_vertices,) — convexity mask
s.directed_edge_directions # (2*n_edges, 2)
s.edge_lengths # (n_edges,)
s.edge_midpoints # (n_edges, 2)
s.edge_normals # (n_edges, 2)
s.ring_types # (n_rings,) — "exterior"|"interior"
s.vertex_angles # (n_vertices,) — interior angles (°)
s.vertex_coordinates # (n_vertices, 2)
s.vertex_normals # (n_vertices, 2) — outward unit normals
# topology — bridges (*_to_*)
s.directed_edge_to_edge # directed_edge → edge
s.directed_edge_to_vertex # directed_edge → [start, end]
s.edge_to_vertex # edge → [start, end]
s.ring_to_part # ring → part
s.vertex_to_incoming_directed_edge # vertex → directed_edge
s.vertex_to_outgoing_directed_edge # vertex → directed_edge
s.vertex_to_neighbor_vertex # vertex → [prev, next]
s.vertex_to_part # vertex → part
s.vertex_to_ring # vertex → ring
# extrusions → Solid
s.linear_extrude(height=3)
s.radial_extrude(axis=y(), pivot=x(5)) # start=0, end=360, segment_count=64
s.path_extrude(path) # fillet_segments=None, min_fillet_radius=None, intermediate_sections=None, strategy=None
# sweep strategies (for path_extrude strategy= parameter)
scale_sweep(end=3) # start=1.0
rotate_sweep(angle=360) # start_angle=0.0
resize_sweep(end_size=[2, 4]) # start_size=None
reverse_sweep(strategy=scale_sweep(3))
# export
s.to_dxf_file("output.dxf")
s.to_html_file("output.html")
s.to_screen()
s.to_svg_file("output.svg")
3D — Solid
from scadpy import *
# primitives
cone(radius=2, height=4) # segment_count=64
cuboid(size=[4, 3, 2])
cylinder(radius=2, height=4) # segment_count=64
polyhedron(vertices=vertices, faces=faces)
sphere(radius=3) # subdivision_count=4
Solid.from_stl("model.stl")
# boolean operations
a = cuboid(size=[4, 3, 2]); b = sphere(radius=2)
a | b # union
a - b # difference
a & b # intersection
a ^ b # symmetric difference
a + b # concat (no merge)
# transforms
a.color(color=RED)
a.convexify() # part_filter=None
a.linear_pattern(counts=4, steps=x(3)) # counts=[nx, ny, nz], steps=[sx, sy, sz]
a.mirror(normal=[1, 0, 0]) # pivot=0
a.pull(distance=1.0) # pivot=0, vertex_filter=None
a.push(distance=1.0) # pivot=0, vertex_filter=None
a.radial_pattern(count=6, axis=z()) # angle=360, pivot=0
a.resize(size=[6, None, None]) # auto=False, pivot=None, vertex_filter=None
a.rotate(angle=30, axis=z()) # pivot=0, vertex_filter=None
a.scale(scale=[2, 1, 0.5]) # pivot=0, vertex_filter=None
a.translate(translation=[1, 0, 0]) # vertex_filter=None
# features
a.bounds # [min_x, min_y, min_z, max_x, max_y, max_z]
a.bounding_box # → Solid (cuboid)
a.centroid # [cx, cy, cz] — geometric centroid
a.is_empty # bool
# topology — coordinates & bridges (*_to_*)
a.triangle_to_vertex # triangle → [v0, v1, v2]
a.vertex_coordinates # (n_vertices, 3)
a.vertex_to_part # vertex → part
# export
a.to_html_file("output.html")
a.to_screen()
a.to_stl_file("output.stl")
Roadmap
- Improve documentation
- Richer topology for Shape and Solid
- Richer transformations for Shape and Solid
- Chamfer and fillet on Solid
- New assembly types:
PointCloud2d,Wire2d,PointCloud3d,Wire3d - Better error messages
- More import/export formats
Development
# Create and activate venv
python3 -m venv .venv
source .venv/bin/activate
# Install with dev dependencies
pip install -e .[dev]
# Run doctests & generate documentation & AI skill file
cd docs && make doctest && make html && make skills
License
See LICENSE.md.
Release files for ScadPy 0.4.36
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| scadpy-0.4.36.tar.gz | 486.5 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| scadpy-0.4.36-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 691.1 kB
Release files / scadpy-0.4.36.tar.gz
| Download URL | scadpy-0.4.36.tar.gz |
|---|---|
| Size | 486.5 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
75e125473ea5e1ee3e3fc59baef85a853558394af831014ed31ed8bb32b5b658
|
|
BLAKE2b-256 checksum How to use checksums |
984b95bac9d02a7dee94fba2da9e74aab974375f951ce495f7f33b603f5d231f
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 24, 2026.
Transparency logRelease files / scadpy-0.4.36-py3-none-any.whl
| Download URL | scadpy-0.4.36-py3-none-any.whl |
|---|---|
| Size | 204.5 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
c9498659b00e255542823dfbdbd056df8b8a7dac353743bcef574a069605d76c
|
|
BLAKE2b-256 checksum How to use checksums |
7335a5bfa06a556292f2a4c66c639a6345d7e59cc3bfad5c073fd0554905d58b
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 24, 2026.
Transparency log