cc-skill-creator

cc-skill-creator

Creates and evaluates Agent Skills following the official specification. Use when the user wants to create a new skill, scaffold a skill directory, evaluate or validate an existing skill, or needs guidance on skill structure and format.

1星標
0分支
更新於 1/23/2026
SKILL.md
readonlyread-only
name
cc-skill-creator
description

Creates and evaluates Agent Skills following the official specification. Use when the user wants to create a new skill, scaffold a skill directory, evaluate or validate an existing skill, or needs guidance on skill structure and format.

version
"1.0"

Skill Creator

This skill helps you create new Agent Skills that follow the official specification.

When to Use

  • User wants to create a new skill
  • User needs to scaffold a skill directory structure
  • User asks about skill format or structure
  • User wants to evaluate or validate an existing skill (e.g., "〜のスキルを評価して")

Steps to Create a New Skill

1. Gather Requirements

Ask the user for:

  • Skill name: Must be lowercase, use hyphens, 1-64 characters (e.g., pdf-processing, code-review)
  • Description: What the skill does and when to use it (1-1024 characters)
  • Optional: License, compatibility requirements, metadata

2. Confirm Installation Location

IMPORTANT: Before creating any files, confirm the installation location with the user.

Default Installation Path: .claude/skills

The installation path varies by tool/environment. Determine a reasonable default based on:

  • Current working directory structure
  • Existing skills directories (e.g., ./skills/, ./.claude/skills/)
  • Project conventions

Then ask the user for confirmation:

スキル「{skill-name}」を以下の場所に作成します:
  {full-path}/{skill-name}/

この場所でよろしいですか?別の場所を指定することもできます。

Only proceed after user confirms the location.

3. Create Directory Structure

Create the skill directory at the confirmed location:

skill-name/
├── SKILL.md          # Required - main skill file
├── references/       # Optional - additional documentation
├── scripts/          # Optional - executable code
└── assets/           # Optional - static resources

4. Create SKILL.md

The SKILL.md file must contain:

  1. YAML Frontmatter (required):

    ---
    name: skill-name
    description: A clear description of what this skill does and when to use it.
    ---
    
  2. Markdown Body (recommended sections):

    • Step-by-step instructions
    • Examples of inputs and outputs
    • Common edge cases

5. Validation Checklist

Before finishing, verify:

  • [ ] name field matches the directory name
  • [ ] name is lowercase with hyphens only, no consecutive hyphens
  • [ ] description clearly explains what the skill does AND when to use it
  • [ ] Main SKILL.md is under 500 lines
  • [ ] File references use relative paths from skill root

Name Constraints

The name field must:

  • Be 1-64 characters
  • Use only lowercase letters, numbers, and hyphens (a-z, 0-9, -)
  • Not start or end with -
  • Not contain consecutive hyphens (--)
  • Match the parent directory name exactly

Valid: pdf-processing, code-review, data-analysis
Invalid: PDF-Processing, -pdf, pdf--processing, pdf-

Example SKILL.md

---
name: example-skill
description: Does X and Y. Use when the user mentions A, B, or C.
license: MIT
metadata:
  author: your-name
  version: "1.0"
---

# Example Skill

Instructions for the agent go here.

## Usage

1. First step
2. Second step

## Examples

Input: ...
Output: ...

Optional Directories

references/

Additional documentation loaded on demand. Keep files focused and small.

scripts/

Executable code (Python, Bash, JavaScript). Should be self-contained with clear error messages.

assets/

Static resources: templates, images, data files, schemas.

Evaluating an Existing Skill

When the user asks to evaluate a skill (e.g., "〜のスキルを評価して"), follow these steps:

1. Read the Specification

First, read skills-spec.md to understand the complete format requirements.

2. Read the Target Skill

Read the target skill's SKILL.md file and any related files.

3. Check Against Specification

Evaluate the skill against these criteria:

Required Checks:

  • [ ] SKILL.md exists in the skill directory
  • [ ] YAML frontmatter is present and valid
  • [ ] name field exists and follows naming rules:
    • 1-64 characters
    • Lowercase letters, numbers, hyphens only
    • No leading/trailing hyphens
    • No consecutive hyphens (--)
    • Matches parent directory name
  • [ ] description field exists (1-1024 characters)
  • [ ] Description explains both WHAT the skill does AND WHEN to use it

Recommended Checks:

  • [ ] SKILL.md is under 500 lines
  • [ ] Body contains step-by-step instructions
  • [ ] Examples are provided
  • [ ] File references use relative paths
  • [ ] Reference files are focused and not too large

Optional Field Checks (if present):

  • [ ] license field is concise
  • [ ] compatibility field is under 500 characters
  • [ ] metadata has string keys and string values
  • [ ] allowed-tools is space-delimited

4. Report Findings

Present the evaluation results to the user:

  1. List all issues found (errors and warnings)
  2. Provide specific suggestions for each issue
  3. Show example corrections where applicable

5. Wait for User Permission

IMPORTANT: Do NOT make any modifications automatically. Always:

  1. Present the evaluation report
  2. Ask the user if they want to apply the suggested fixes
  3. Only proceed with modifications after explicit user approval

Example response format:

## 評価結果

### エラー (必須修正)
1. `name` フィールドが...

### 警告 (推奨修正)
1. `description` が...

### 修正案
[具体的な修正内容]

これらの修正を適用しますか?

Reference

For the complete specification, see skills-spec.md.

You Might Also Like

Related Skills

coding-agent

coding-agent

179Kdev-codegen

Run Codex CLI, Claude Code, OpenCode, or Pi Coding Agent via background process for programmatic control.

openclaw avataropenclaw
獲取
add-uint-support

add-uint-support

97Kdev-codegen

Add unsigned integer (uint) type support to PyTorch operators by updating AT_DISPATCH macros. Use when adding support for uint16, uint32, uint64 types to operators, kernels, or when user mentions enabling unsigned types, barebones unsigned types, or uint support.

pytorch avatarpytorch
獲取
at-dispatch-v2

at-dispatch-v2

97Kdev-codegen

Convert PyTorch AT_DISPATCH macros to AT_DISPATCH_V2 format in ATen C++ code. Use when porting AT_DISPATCH_ALL_TYPES_AND*, AT_DISPATCH_FLOATING_TYPES*, or other dispatch macros to the new v2 API. For ATen kernel files, CUDA kernels, and native operator implementations.

pytorch avatarpytorch
獲取
skill-writer

skill-writer

97Kdev-codegen

Guide users through creating Agent Skills for Claude Code. Use when the user wants to create, write, author, or design a new Skill, or needs help with SKILL.md files, frontmatter, or skill structure.

pytorch avatarpytorch
獲取

Implements JavaScript classes in C++ using JavaScriptCore. Use when creating new JS classes with C++ bindings, prototypes, or constructors.

oven-sh avataroven-sh
獲取

Creates JavaScript classes using Bun's Zig bindings generator (.classes.ts). Use when implementing new JS APIs in Zig with JSC integration.

oven-sh avataroven-sh
獲取