Computational Design Skills
A Claude Code Skills Plugin for Computational Designers
18 interconnected skills covering computational geometry, parametric modeling, generative design, structural computation, environmental simulation, facade engineering, digital fabrication, BIM scripting, and machine learning for AEC. References 50+ pioneers, 30+ tools, 6+ analysis engines, and hundreds of numeric benchmarks. Includes 7 Python calculators for geometry, structural checking, solar analysis, panel optimization, mesh analysis, material estimation, and fabrication costing.
35,000+ lines | 75+ files | 18 skills | 7 Python calculators
Table of Contents
- Quick Start
- How It Works
- Skills Reference
- Foundation
- Core Design
- Analysis & Simulation
- Fabrication & Production
- BIM & Integration
- Optimization & Intelligence
- Computation
- Computational Design Calculator
- Knowledge Base
- Architecture
- Example Prompts
- Stats
- Author & License
Quick Start
Option 1: Use as a Plugin
claude --plugin-dir "/path/to/Computational Design Skills"
Option 2: Copy to Your Project
cp -r skills/ your-project/.claude/skills/
Once installed, just ask Claude anything about computational design, parametric modeling, generative algorithms, structural analysis, facade engineering, or digital fabrication. The system activates automatically - no slash commands required for most tasks.
How It Works
Computational Design Skills uses a progressive disclosure architecture to stay lean while encoding deep technical knowledge:
User asks anything about computational design
|
v
cd-foundations (auto-fires, invisible to user)
- Applies baseline knowledge from 50+ frameworks and pioneers
- Routes to specialized skills as needed
|
v
Specialized skill loads (e.g., parametric-modeling, facade-computation)
- Follows structured workflow
- Loads deep reference files on demand
|
v
cd-calculator provides precise computations when needed
- 7 Python scripts for geometry, structural checking, solar
analysis, panel optimization, mesh analysis, material
estimation, and fabrication costing
Three-Layer Context Management
| Layer | What Loads | When | Size |
|---|
| Layer 0 | Skill names + descriptions only | Always in context | ~16KB |
| Layer 1 | Full SKILL.md body | On skill invocation | ~350-900 lines each |
| Layer 2 | Reference files, templates, scripts | Only when explicitly needed | Varies |
This keeps the context window efficient. Claude never loads knowledge it doesn't need for the current task.
Skills Reference
1. Foundation (Auto-Activated)
cd-foundations
Invocation: Automatic (invisible to user) - Claude-only, not user-invocable
The foundation layer. Auto-fires on any computational design context and provides:
- Pioneer Quick-Reference Matrix - 25+ pioneers with key contributions, methods, and computational paradigms (Frei Otto, Greg Lynn, Patrik Schumacher, Mark Burry, Achim Menges, Robert Aish, and more)
- Computational Design Taxonomy - Classification of 12 computational paradigms: parametric, generative, algorithmic, evolutionary, agent-based, physics-based, data-driven, topological, mesh-based, voxel-based, graph-based, and ML-driven
- Geometric Primitives & Operations - NURBS, B-splines, subdivision surfaces, mesh operations, boolean operations, convex hulls, Voronoi, Delaunay, offset curves, lofting, sweeping
- Tool Ecosystem Map - 30+ tools cross-referenced: Rhino, Grasshopper, Dynamo, Revit API, Houdini, Processing, OpenSCAD, Blender Geometry Nodes, Unity, Unreal, TouchDesigner, FreeCAD
- Anti-Pattern Catalog - Common computational design failures: over-parameterization, mesh quality collapse, solver divergence, fabrication-ignorant design
- Skill Router - Decision tree that activates specialized skills based on user context
Reference files (loaded on demand):
pioneers-and-movements.md - Full treatment of all pioneers and computational movements
core-concepts.md - Complete mathematical and geometric reference
tools-ecosystem.md - Detailed tool comparison, API surfaces, interoperability matrix
learning-pathways.md - Beginner-to-expert learning roadmaps by tool and domain
2. Core Design
parametric-modeling
Invocation: /parametric-modeling or auto-activates on parametric/parameter-driven context
Parametric design methodology and pattern authoring:
- Parametric thinking framework - From design intent to parameter identification to dependency graph (DAG) construction, with lazy vs. eager evaluation
- Grasshopper component mapping - 200+ components organized by category with input/output signatures, plus Dynamo node equivalents for cross-platform workflows
- Data tree manipulation - Graft, flatten, flip, path mapper, branching trees, and history tree patterns for complex parametric definitions
- Parametric facade systems - Attractor-based, field-driven, image-sampled, data-mapped panel variation with design space exploration
- Performance feedback loops - Structural, environmental, and cost metrics driving parameter adjustment in real time
Reference files: grasshopper-patterns.md, dynamo-patterns.md, data-structures.md
generative-design
Invocation: /generative-design or auto-activates on generative/evolutionary/optimization context
Generative and evolutionary design systems:
- Generative algorithms taxonomy - L-systems, cellular automata, agent-based modeling, shape grammars, space syntax generation, wave function collapse, graph grammars
- Evolutionary optimization - Genetic algorithms (GA), NSGA-II for multi-objective, CMA-ES, differential evolution, particle swarm optimization with convergence diagnostics
- Fitness function design - Single-objective, multi-objective, constraint handling, penalty methods, feasibility-first ranking
- Generative floorplans & massing - Adjacency-based generation, stacking diagrams, voxel-based massing, solar envelope generation, view corridor preservation
- Genotype-phenotype mapping - Direct encoding, indirect (developmental) encoding, grammatical encoding, neural network encoding
Reference files: evolutionary-algorithms.md, fitness-functions.md
computational-geometry
Invocation: /computational-geometry or auto-activates on geometry/NURBS/mesh/surface context
Core computational geometry methods for AEC:
- Curve & surface mathematics - Bezier, B-spline, NURBS with knot vectors, T-splines, subdivision surfaces (Catmull-Clark, Loop, Doo-Sabin), parametric domains, trimming
- Solid modeling - CSG (constructive solid geometry), B-rep (boundary representation), voxel grids, implicit surfaces, signed distance fields
- Intersection & boolean operations - Curve-curve, curve-surface, surface-surface, ray-mesh intersection, union/difference/intersection with robustness strategies
- Voronoi, Delaunay & convex hull - 2D/3D Voronoi diagrams, weighted/centroidal Voronoi, Delaunay triangulation, Graham scan, Quickhull, Minkowski sum/difference
- Spatial indexing & robustness - BVH, k-d trees, octrees, R-trees, floating-point pitfalls, exact arithmetic, snap rounding, geometric predicates
Reference files: nurbs-reference.md, mesh-fundamentals.md, tessellation-methods.md
algorithmic-patterns
Invocation: /algorithmic-patterns or auto-activates on pattern/tiling/tessellation context
Algorithmic pattern generation and tiling systems:
- Tessellation library - Regular (3), semi-regular (8), demi-regular (14), Penrose (2), Ammann-Beenker, pinwheel, aperiodic Hat monotile (2023), Spectre tile
- Islamic geometry construction - Compass-and-straightedge methods, star polygon generation, girih tiles (5 types), muqarnas vaulting algorithms
- Fractal and recursive patterns - Sierpinski, Koch, Hilbert curve, L-system grammars for botanical modeling, reaction-diffusion (Gray-Scott, Turing)
- Attractor fields & packing - Point/curve/surface attractors, inverse-distance weighting, Gaussian falloff, circle packing (Apollonius), sphere packing, bin packing for nesting
- Weaving and interlocking - Warp-weft logic, triaxial weaving, reciprocal frame structures, kagome lattice
Reference files: pattern-catalog.md, biological-computation.md
3. Analysis & Simulation
structural-computation
Invocation: /structural-computation or auto-activates on structural/FEA/load/stress context
Structural analysis and form-finding for AEC:
- Form-finding methods - Force density method, dynamic relaxation, thrust network analysis, particle-spring systems, hanging chain models, graphic statics
- Finite element fundamentals - Element types (bar, beam, shell, solid), shape functions, stiffness matrix assembly, boundary conditions, load cases
- Shell & tensile structures - Gaussian curvature analysis, funicular shells, gridshells, membrane form-finding, cable-net analysis, minimal surfaces, patterning and compensation
- Structural optimization - Topology optimization (SIMP), size and shape optimization, compliance minimization, stress constraints, load path visualization
- Quick checks - Span-to-depth ratios for 12 structural systems, approximate member sizing, deflection limits (L/250, L/360), steel/concrete/timber rule-of-thumb tables
Reference files: form-finding-methods.md, fea-fundamentals.md, topology-optimization.md
environmental-simulation
Invocation: /environmental-simulation or auto-activates on solar/wind/daylight/thermal/acoustic context
Environmental performance simulation for buildings and urban contexts:
- Solar & daylight - Sun path diagrams, shadow studies, radiation maps (kWh/m2), daylight factor, spatial daylight autonomy (sDA300/50%), annual sunlight exposure (ASE1000/250h), Radiance/DAYSIM parameters
- Wind & thermal - CFD fundamentals (RANS, LES), Lawson comfort criteria, natural ventilation potential, U-value calculation, adaptive comfort (EN 15251), PMV/PPD
- Acoustic modeling - Room acoustics (RT60, EDT, C80), urban noise mapping, STC, speech intelligibility (STI), noise barrier design
- Energy & carbon - EnergyPlus/IES VE inputs, HVAC system selection, operational carbon, embodied carbon (kgCO2e/m2), whole-life carbon assessment
- Microclimate - UTCI outdoor thermal comfort, mean radiant temperature, sky view factor, urban heat island quantification
Reference files: daylight-standards.md, cfd-basics.md, energy-modeling.md
4. Fabrication & Production
facade-computation
Invocation: /facade-computation or auto-activates on facade/panel/curtain wall/cladding context
Facade engineering and panelization computation:
- Panelization strategies - Planar quad, planar triangle, ruled surface, single-curved developable, double-curved formed, cold-bent glass, mold-cast, 3D-printed
- Rationalization algorithms - Planar remeshing (PQ mesh), conical mesh, circular mesh, edge offset mesh, curvature-based segmentation, panel clustering for mold reuse
- Curtain wall systems - Stick, unitized, point-fixed, cable-net, structural glazing, double-skin, media facade, kinetic facade