Ni Jingzhe 2029beedea scene + exporters: a Part is drawn, exported and measured from its one body (no per-body branches)
DECISION: every consumer of a Part reads `part.body` again. The code that treated a multi-body Part as a
set of child solid nodes, a Compound view or a list of named / coloured STEP solids is removed.

WHY - THE DISCUSSION (owner and assistant, 2026-11-10; full account in design-docs/part-holds-one-solid.md)
The multi-body Part was judged a wrong design (it was made so the editor could hold several solids
with no assembly constraint, and it makes assembly hard); the agreed fix is Part = one solid, and the
editor's several solids become loose preview values (commit "runtime + preview"). The audit of 2.2.8
showed what a second code path per consumer costs: each consumer had to branch on `multi_body`, and some
did not (explode, the legacy scene compiler), which is how the release came to have crashes. With the
invariant enforced in the Part constructor these branches can never run, so they are deleted rather than
kept as dead code.

WHAT CHANGES
- scene/product_scene.py: the package scene has one GLB and one entity document per Part again;
  _body_geometries and the `node/<part>/body:<id>` child solid nodes are removed; _part_geometry no
  longer refuses; _appearance_properties reads the part's one appearance row.
  (The package-scene node tree of a one-body part is what it always was.)
- exporter/step.py: the STEP product's shape is `part.body`; the body colour overrides the material colour
  as before; the `bodies` list of named / coloured solids is no longer built for a Part. The STEP writer's
  generic support for named solids (io::StepBody in C++) is left in place: it is the writer's ability, it
  has its own tests, and removing it would be a separate decision.
- exporter/mjcf.py: the collision / visual mesh of a Part is `part.body`.
- scene/prewarm.py, exporter/_surface_mesh.py, build/definition.py: only the comments that said "every
  body of a multi-body part" (the loops over `part.bodies` stay: it has one element).

Checked: test_product_exporter, test_product_packages, test_scene_preview, test_scene_compiler and
test_notebook_runtime pass. test_body_appearance fails on its multi-body fixtures; the tests are updated in a
later commit of the series.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-10 03:53:50 +08:00
2026-11-09 14:02:31 +08:00
2026-11-09 14:02:31 +08:00
2026-09-06 18:42:33 +08:00
2026-11-09 14:02:31 +08:00
2026-11-09 14:02:31 +08:00

SimpleCADAPI repository cover

SimpleCADAPI

Parametric CAD built for agents.
Efficient · Low-overhead · Agent-native

中文 · Changelog · Skill design · Paper

Start here. Paste this line to your coding agent (Claude Code, Codex, OpenCode, …):

Read physpace.dev/sca/landing.md and landing me on this sdk.

It installs the SDK, installs the skill for itself, opens a real example in the editor, and shows you how to work with it.


The idea

CAD software was made for a person with a mouse. An agent has no mouse. It has code, a context window and a token bill. SimpleCADAPI starts from that.

Efficient More first attempts come out right, and fewer tokens are spent getting there. Change one parameter and only the cells that depend on it run again.
Low-overhead One pip install: the geometry engine is native C++ on OpenCASCADE inside the wheel, with no CadQuery or OCP stack. A functional API with no hidden state, so every step is a cell that can be cached, checked and replayed. A skill an agent reads in one batch, not twenty.
Agent-native Errors say what happened, give the measured numbers and say how to fix it. Geometry is selected by what it is (ql, stable scad.ref), not by an index that moves. Every call is recorded as a replayable graph. The editor writes what you click into the notebook, so your agent reads your selection from disk.

The evidence

Against CadQuery. Same model, same prompt, same 200 real parts, one attempt each. A frozen, independent scorer compares each result with the original.

Same model and prompt on 200 parts: SimpleCADAPI delivers 3.5 times fewer badly wrong models than CadQuery

DeepSeek-V4.1-Flash, SimpleCADAPI 2.2.7b1 vs CadQuery 2.8.0, 200 open-hardware parts from 27 projects. "Badly wrong" is a delivered model whose surface F1 at 1 mm is below 99.99 %. Disclosed in the report: the SimpleCADAPI run also had a STEP-reconstruction workflow guide, the CadQuery run a general third-party CadQuery skill; the parts are mostly simple (median 19 faces).

Against a mature industrial CAD. Tokens spent per task; the industrial CAD is driven through its MCP server.

Tokens spent per task: SimpleCADAPI uses up to 6.5 times fewer than a mature industrial CAD through its MCP

What it builds

Two six-axis arms, each an assembly of part notebooks composed with scad.use. Every joint angle and link length is a named parameter: drag a slider and the model follows.

The SG90 six-axis arm in the SimpleCADAPI editor: notebook cells and joint sliders on the left, the live assembly on the right

SG90 educational arm · 34 parts, six revolute joints solved as constraints, 3D-printable links, reserved envelopes for the electronics. sg90_arm_6dof.py

The desktop six-axis arm with a gripper in the SimpleCADAPI editor

Desktop six-axis arm with gripper · 34 parts. The joint motor is a purchased part with its real mounting interface (34 × 20 × 19.6 mm body, six M2 holes on a Ø10.5 circle), and the links carry the matching pockets, screw channels and plug windows. arm6.py

The editor, sca edit, is code-first: cells on the left, the live model on the right, a slider for every parameter. Click a solid, face, edge or vertex and it writes a selection into the notebook. Your agent reads what you picked from the file, so "chamfer these two edges" needs no ids. The editor and sca run share one runtime and one per-cell cache; on the six-axis arm a joint angle already seen re-renders in 0.6 s, a new one in 2.7 s (2.2.7).

The skill

The skill is how an agent learns the SDK. It is a router and a library of small pages, written so the agent reads exactly what a step needs, in one parallel batch.

What it does — seven routes, each a complete workflow with its own checks:

Ask Route
Model one physical part single-part modeling: requirements → plan → build → verify
Build a multi-part product or mechanism assembly: connectors, constraints, kinematics, standard parts
Rebuild an editable model from a STEP file STEP reconstruction with stable entity ids and acceptance gates
Rebuild from a mesh, point cloud or photos scan reconstruction
Rebuild from a vector-PDF drawing drawing reconstruction
Produce a GB engineering drawing (DXF) engineering drawing from validated geometry
Hand the model to someone else export and translation: STEP AP242, STL, FreeCAD, MJCF

How it is organized

SKILL.md                  router: pick one route, read its batch (≤ 130 lines)
workflows/                one lifecycle per route; roles and stages load when they start
domains/                  what the SDK offers for a kind of task
discipline/               rules and invariants: core rules, feature order, verifiers, failure and repair
docs/api · stdlib · core  400+ generated signature pages; a method with no page does not exist
  • Small reading budget. The first batch of any route is at most 1,200 lines. Every page has one job, and every rule has one home. A build check fails the skill if a page breaks the contract.
  • Two modes. Every workflow runs interactive (a person can answer) or unattended (nobody will). The mode changes who decides, never what is checked.
  • One source, seven harnesses. The same pages compile for Claude Code, Codex, OpenCode, ZCode, OMP, DSH and Kimi Code.
sca skill targets            # the harnesses
sca skill install            # compile for yours and install

Plugins

Modeling is the SDK's job. Checking, simulating and translating a finished model belong in a plugin, and a plugin is a repository:

sca-addon.toml     machine-checked facts: name, version, which SDK versions, how to provision
skill/SKILL.md     everything an agent reads about it
...                its own code, binaries and docs, in any language
sca addon add owner/repo          # from GitHub; owner/repo@v1.2.0 pins a tag
sca addon add ./my-addon          # a local checkout
sca addon list                    # what is installed, and whether its runtime works
sca addon update                  # re-fetch; code is swapped, runtime state is kept
  • The agent learns it on install. The addon's skill lands beside the SDK's as sca-<name>, so the agent gets the capability and the instructions together.
  • Isolated by design. A plugin reads .scadpkg product packages across a process boundary and runs in its own environment, so plugin dependencies never touch the one that models geometry.
  • Already in use. Connector verification (bolted, pinned, press-fit, welded and other joints, checked and then run through FEM) and FEM mesh preparation (fasteners removed, their holes filled, closure proven) are plugins. The AP242 → Gmsh → CalculiX and MJCF → MuJoCo paths below run off the same package.
Bracket FEM von Mises
L-bracket static FEM
CalculiX von Mises on the Gmsh mesh of the AP242 export; faces are chosen by interface.* tags that survive model changes
Four-bar links crashing through each other in MuJoCo
Four-bar linkage, first pass
MuJoCo shows the links crossing: the assembly is wrong
Corrected four-bar assembly in MuJoCo
After the collision pass
the same crank cycle sweeps clean

Install

pip install "simplecadapi[editor]"      # the SDK and the editor
sca example u_link                       # a real workspace to open
sca edit u_link                          # the editor
sca skill install                        # the skill for your agent

Python 3.10–3.13 on Linux, macOS and Windows. uv tool install "simplecadapi[editor]" puts sca on your PATH without a virtual environment. sca run model.py runs a notebook headless; sca export writes STEP, STL and OBJ.

Research artifact

This repository is an artifact of CADIR: A Cross-Backend Editable Intermediate Representation for Agentic CAD Generation.

Apache-2.0. Validate generated models before production use.

S
Description
No description provided
Readme Apache-2.0 77 MiB
Languages
Python 58.3%
C++ 30.5%
TypeScript 11%
CMake 0.1%