CADIR: A Cross-Backend Editable Intermediate Representation for Agentic CAD Generation

arXiv 2026 57.4 method

TLDR

CADIR is an agent-friendly intermediate representation for CAD generation, enabling cross-backend editable feature histories via construction graphs and geometric signature matching.

Reasoning

The paper introduces a novel intermediate representation that addresses cross-backend editability and construction history preservation, a significant strength. However, the abstract is cut off, limiting full assessment of experimental results, and the approach is tied to the OCCT kernel, which may restrict generality.

Read-first score

Read-first score 57.4, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 102.

Recency 8%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Topical relevance 42%

63.7

Uses existing LLM keyword relevance scores normalized to 0-100. AI for CAD,computer-aided design,neural CAD,generative CAD,parametric CAD,B-Rep,boundary representation,constructive solid geometry,CSG,sketch extrusion,CAD generation,CAD reconstruction,text-to-CAD,image-to-CAD,point cloud to CAD,CAD program

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=experiment,metric

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 42.

Keyword Scores

computer-aided design

CAD generation

parametric CAD

CAD reconstruction

CAD program

generative CAD

text-to-CAD

image-to-CAD

AI for CAD

B-Rep

boundary representation

sketch extrusion

neural CAD

constructive solid geometry

CSG

point cloud to CAD

Deep Analysis

Innovations

Cross-backend editable intermediate representation (CADIR) that records explicit construction graphs with parameter dependencies, constraints, and topology selections
Geometric Signature Matching for identifying corresponding edges/faces across parameter changes and backend differences, enabling native editable feature history reconstruction in FreeCAD, SolidWorks, and Fusion 360
Construction-graph retrieval method for text and image queries supporting full-graph and subgraph retrieval to leverage complete models and substructures for generation

Methodology

CADIR is built on the OCCT geometry kernel via OCP, recording modeling operations, parameter dependencies, constraints, and topology selections in a construction graph during execution. Geometric Signature Matching identifies corresponding edges and faces across backends, allowing adapters to reconstruct native editable feature histories. A construction-graph retrieval method is proposed for text and image queries, supporting both full-graph and subgraph retrieval to aid agents in reusing models and substructures.

Key Results

CADIR achieves higher geometric fidelity and execution reliability than existing CAD representations; construction-graph retrieval further improves model generation quality; and cross-backend editing enables reliable model reconstruction and post-reconstruction editing across FreeCAD, SolidWorks, and Fusion 360.

Resources

Paper