Draft SKILL.md content and run with-skill versus without-skill trigger tests.
npx skills add https://github.com/GameDevFox/neko --skill skill-creator-gamedevfox
Please help me install this Agent Skill. Skill: skill-creator Source: https://github.com/GameDevFox/neko/tree/main/claude/skills/skill-creator Command: npx skills add https://github.com/GameDevFox/neko --skill skill-creator-gamedevfox
This Skill solves the problem of designing effective Claude skills by turning an idea into a working draft and then iterating with structured test prompts and quantitative evaluation.
SKILL.md frontmatter description to improve when Claude actually decides to use the skill.Use the skill creator to turn your workflow into a new Claude skill: “I want to create a skill that helps with X; here’s what success looks like and how I expect Claude to behave—please draft the SKILL.md, propose eval prompts, and help me iterate with benchmarks.”
High-intent search queries and answers about installing and using this skill.
To create and evaluate Claude skills, draft the SKILL.md content, generate test prompts, run with-skill vs without-skill trigger tests, and iteratively refine the skill based on qualitative and quantitative benchmark results.
Trigger optimization for LLM instructions works by refining the SKILL.md frontmatter description to improve how accurately Claude decides to use the skill, directly addressing under-triggering or over-triggering issues during evaluation.
The best way to debug incorrect skill triggering is to run structured with-skill versus without-skill evaluation tests, use grader-based expectation scoring, and apply safe improvement guidance to refine the skill description.
Yes, you can benchmark Claude skill performance by executing evaluation prompts, aggregating benchmark metrics, and using grader-based scoring to measure performance variance and ensure better handling of edge cases.
You do not need existing test prompts for skill creation; you can draft a new skill from an outlined workflow and the system will generate realistic test and interview questions to start the iterative loop.
You should avoid test-driven skill iteration if your workflow cannot support a structured loop of intent capture, eval execution, and grader-based expectation scoring, or if it requires introducing unsafe and deceptive behavior.
daikichiba9511Automate the creation, evaluation, and iteration of Claude skills with SKILL.md.
yahav123147Create and iterate SKILL.md structures with trigger evaluations and grading evidence.
hackutdCreate, test, evaluate, and refine Claude skills through a structured workflow.
zhbonerDraft SKILL.md files and refine them through prompt-based test evaluations.