forgecad-grade-model

Grades ForgeCAD models against requirements using renders, inspections, and a 0-10 rubric.

927|103|Updated Feb 8, 2026
One-click install
npx skills add https://github.com/KoStard/forgecad-public-kit --skill forgecad-grade-model
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: forgecad-grade-model
Source: https://github.com/KoStard/forgecad-public-kit/tree/main/skills/forgecad-grade-model
Command: npx skills add https://github.com/KoStard/forgecad-public-kit --skill forgecad-grade-model

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Evaluating whether a code-generated CAD model actually satisfies its brief is subjective and error-prone. This Skill enforces an evidence-based grading workflow so ForgeCAD models are scored against the stated requirement with rendered views, inspections, and explicit score caps instead of guesswork.

Core Features & Use Cases

  • Requirement Checklist Extraction: Converts a brief, prompt, or acceptance criteria into must-have and nice-to-have items, marking unverifiable items as Unknown.
  • Evidence-Based Scoring: Runs the model, renders iso/front/right/top views, inspects hidden internals via the forgecad-inspect-model skill, and applies a weighted rubric with hard score caps.
  • Structured Grade Report: Produces a 0-10 score with a requirement checklist table, evidence citations, applied caps, and the highest-leverage fixes.
  • Use Case: After generating a multi-part assembly in ForgeCAD, ask for a grade to verify the component model, connector mates, and mechanical plausibility before delivering it as a prototype.

Quick Start

Use the forgecad-grade-model skill to grade my model file bracket.forge.js against the original brief and give me a 0-10 score with evidence.

Frequently Asked Questions about forgecad-grade-model

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I grade a ForgeCAD model against a design brief?

Run the model with forgecad run, render iso, front, right, and top views, visually inspect every PNG, then score against a rubric covering requirement fit, geometry, mechanical plausibility, validation, and code quality. The skill outputs a 0-10 score with evidence and caps.

What evidence is required before scoring a CAD model?

The model must execute successfully and be rendered from at least four views that are actually opened and inspected. Scoring from command output alone caps the score at 5, and uninspected internals cap it at 7 when fit or walls matter.

Does the grading skill modify or repair the model?

No, it never edits the model unless the user explicitly requests repairs. If repairs are requested, it records the baseline grade first, applies changes, and then re-grades the updated model.

Why is my ForgeCAD assembly score capped at 7?

Multi-part assemblies that violate the ForgeCAD component model are capped at 7, or 6 when the violation makes the assembly brittle. Violations include sibling imports, assembly-space coordinates inside parts, structural translate placement, and missing connector mates.

What happens if the model fails to execute during grading?

If the model does not execute, grading stops and the score is capped at 2. A model that executes but cannot be rendered is capped at 4, since no visual evidence can be gathered.