Catalog Guide

Use this page when you are still deciding which packaged problem to start with. The generated catalog under Problem Catalog is the complete inventory; this guide is the curated entry point.

Role Labels Used In This Guide

• Canonical benchmark: the recommended representative benchmark for a family when you want a stable comparison target.

• Example: the lightest or clearest teaching entry point for learning the family API.

• Convenience demo: a useful integration showcase that is intentionally narrower than the family’s deepest benchmark surface.

Family Start Table

Family	Recommended starting point	Canonical benchmark	Example or demo
Text	Start with Peanut Shelling without Electricity when you want a citation-backed ideation brief with several related variants nearby.	Peanut Shelling without Electricity	Drinking Fountain for a Public Park
Decision	Start with Decision Problem - Student Laptop Design Under Choice-Based Demand for the clearest single-problem introduction to structured alternatives and evaluations.	Decision Problem - Student Laptop Design Under Choice-Based Demand	Decision Problem - MSEval Safety Helmet (Lightweight)
Optimization	Start with Pill capsule minimum-area problem if you want the lightest local optimization loop before moving to larger benchmarks.	Planar truss span structural optimization (mass minimization)	Pill capsule minimum-area problem
Grammar	Start with Planar truss span grammar problem for a representative sequential constructive benchmark with explicit state transitions.	Planar truss span grammar problem	IoT home cooling-system co-design grammar problem
MCP	Start with Build123d scripted mounting bracket design problem; it is both the family benchmark and the clearest integration demo today.	Build123d scripted mounting bracket design problem	Build123d scripted mounting bracket design problem

Recommended Use By Intent

Your goal	Start family	Why
Benchmark open-ended ideation or prompt framing	Text	Text problems package the benchmark brief itself and keep the research prompt visible.
Compare explicit alternatives against criteria	Decision	Decision problems expose candidate spaces and reusable evaluation helpers directly.
Measure algorithmic performance under objectives and constraints	Optimization	Optimization problems expose typed bounds and packaged baseline solve routines.
Study constructive action sequences and transition structure	Grammar	Grammar problems make the state/action model explicit for trace-oriented analysis.
Exercise an external tool runtime through a packaged problem shell	MCP	MCP problems wrap upstream tool servers as a first-class family.

What To Read Next

• Use Problems for the conceptual family overview.

• Use Problem Catalog for the full generated inventory once you know the family.

• Use Downstream Metadata Contract when you want to carry packaged-problem metadata into experiments or analysis.