Question Creator FAQs

Question 1

What difficulty levels and categories are supported?

Accepted Answer

The skill supports four difficulty tiers (base, advanced, final, and final+) and four primary evaluation categories: Code, Mathematics, Logic, and Comprehensive capabilities.

Question 2

How does this skill improve my benchmarking workflow?

Accepted Answer

It eliminates manual errors by enforcing sequential numbering, automated directory naming, and strict metadata validation. It also integrates domain-specific scoring metrics (like code quality or logical accuracy) automatically based on the category you choose.

Question 3

What specific files does the skill generate?

Accepted Answer

For every new question, the skill generates a standardized package including a README.md for human reading, a meta.yaml for machine processing, a prompt.md for the model input, and a reference.md containing the ground truth answer.

Question 4

What does the Question Creator skill do?

Accepted Answer

The Question Creator skill automates the generation of standardized benchmark test cases for the CAC evaluation system. it handles directory structuring, metadata configuration (meta.yaml), and the creation of prompt and reference files across multiple domains like coding, mathematics, and logic.

Question 5

When should I use this Claude Code skill?

Accepted Answer

Use this skill whenever you need to add new evaluation items to your test suite. It is triggered by commands like 'add question', 'create question', or 'new question', helping you maintain a consistent repository structure without manual formatting.

Question Creator

Question Creator

Key Features

Use Cases

Key Features

Use Cases