From core-skills
Scaffold new skills for this repository with consistent folder layout, frontmatter, agent metadata, and optional references, assets, and scripts. Use when the user wants to create another skill in this repo, bootstrap a skill skeleton, normalize a new skill to the local repo conventions, or generate editable starter files instead of writing each skill from scratch.
How this skill is triggered — by the user, by Claude, or both
Slash command
/core-skills:skill-scaffoldThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Create new skills in this repository by following [../../AGENTS.md](../../AGENTS.md) and using the bundled scaffold script.
Create new skills in this repository by following ../../AGENTS.md and using the bundled scaffold script.
references, assets, scripts.Run the script with the smallest useful set of resources.
Example:
pwsh -File .\skills\skill-scaffold\scripts\new-skill.ps1 `
-SkillName "sql-coach" `
-Description "Coach SQL learning and query practice. Use when the user wants step-by-step SQL help, exercises, or review." `
-DisplayName "SQL Coach" `
-ShortDescription "Coaching-first SQL tutor." `
-DefaultPrompt "Teach SQL through hints, review, and deliberate practice." `
-IncludeReferences
description so it clearly triggers on the right requests.SKILL.md procedural and concise.references/.The scaffold is intentionally thin. Replace placeholders rather than expanding them into generic filler.
At minimum, fill in:
SKILL.md frontmatter descriptionagents/openai.yamlSKILL.mdreferences/ when the skill needs mode-specific or domain-specific detail.assets/ when the skill needs reusable templates or output files.scripts/ when a helper should be deterministic and reused.Read references/repo-rules.md when you need the local conventions summarized. Reuse the templates in assets/templates rather than rewriting the same starter files.
npx claudepluginhub bakersean168/skills --plugin core-skillsCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.