Claude Skill Optimizer FAQs

Question 1

Can I run this on a whole repository at once?

Accepted Answer

Yes, the tool includes an 'audit' command specifically designed to batch-process entire directories, providing a status report on every SKILL.md file found.

Question 2

What content is typically moved to the 'extended' layer?

Accepted Answer

The optimizer recommends moving full-length examples (over 20 lines), large templates, detailed checklists, and decorative elements to the extended layer to keep the primary skill file under 300 lines.

Question 3

Will optimizing a skill make it less effective?

Accepted Answer

No. The optimizer ensures that the most critical 20% of information—the logic used in 80% of cases—remains in the core file, while secondary information is preserved in an extended directory for on-demand access.

Question 4

How does this skill reduce token usage?

Accepted Answer

It implements a layered strategy that identifies core knowledge while moving large examples, templates, and complex ASCII diagrams to external files that are only loaded when necessary.

Question 5

What is a 'Token Efficiency Score'?

Accepted Answer

It is a performance metric calculated by the ratio of core content lines to total lines, helping you objectively identify which skills are 'bloated' and require refactoring.

Claude Skill Optimizer

概要

主な機能

ユースケース

Claude Skill Optimizer

概要

主な機能

ユースケース