Help us improve
Share bugs, ideas, or general feedback.
From skill-maker
Packages a skill folder for distribution: generates Claude.ai zip file, Claude Code plugin scaffold, or both. Validates skill, scans project, suggests output paths.
npx claudepluginhub vcode-sh/vibe-tools --plugin skill-makerHow this command is triggered — by the user, by Claude, or both
Slash command
/skill-maker:package path to skill folderThe summary Claude sees in its command listing — used to decide when to auto-load this command
# Package Skill for Distribution Package an existing skill for distribution on Claude.ai, Claude Code, or both. ## Input The user provides: `$ARGUMENTS` If `$ARGUMENTS` is a path, use it directly. If `$ARGUMENTS` is empty, ask: "Which skill should I package? Provide the path to the skill folder." ## Step 1: Pre-Package Validation Run quick validation first (same as `/skill-maker:validate`). If critical issues found, warn the user and offer to fix before packaging. ## Step 2: Choose Target Ask the user: ## Step 3: Project Scan & Output Location Before packaging, scan the project t...
/create-skillCreates a new Claude Skill with YAML frontmatter in personal or project directory. Validates inputs, checks existence, analyzes for bundled resources, and adds scripts/, references/, or assets/ directories as needed.
/install-skillCreates a skill from any source (URL, repo, file) and packages it for a target LLM platform like Claude, OpenAI, or Gemini.
/create-skillGuides interactive creation of a new Claude Code skill via agent workflow, generating validated SKILL.md with requirements, specs, manifest, and analysis.
/create-skillCreates a new Claude Code skill directory from skill name, description/purpose, and optional documentation URLs. Generates SKILL.md with YAML frontmatter, quick start guide, critical rules, and known issues prevention.
/add-skillCreates a new Skill folder with SKILL.md and supporting directories in the current plugin directory or specified marketplace plugin.
/packagePackages Claude Desktop framework files for a PROFILE (e.g., CREATIVE, DEVELOPER) or default, lists generated paths alphabetically, and updates local cache.
Share bugs, ideas, or general feedback.
Package an existing skill for distribution on Claude.ai, Claude Code, or both.
The user provides: $ARGUMENTS
If $ARGUMENTS is a path, use it directly.
If $ARGUMENTS is empty, ask: "Which skill should I package? Provide the path to the skill folder."
Run quick validation first (same as /skill-maker:validate).
If critical issues found, warn the user and offer to fix before packaging.
Ask the user:
Package target:
1. Claude.ai (zip file for upload)
2. Claude Code plugin (plugin scaffold)
3. Both
Before packaging, scan the project to suggest the best output location.
Scan the working directory and surrounding project structure:
package.json, pyproject.toml, composer.json, or similar manifest (if present)**/skills/, **/*-plugin/, **/.claude-plugin/**/*.zip matching skill patternssrc/, dist/, build/, output/, packages/Present a summary of what was found and suggest locations:
PROJECT SCAN: [project name or cwd]
═══════════════════════════════════
[If relevant structures found:]
Detected:
- Skills directory: ./skills/ (contains X skills)
- Plugin directory: ./plugins/ (contains Y plugins)
- Existing packages: ./dist/skill-name.zip
[If nothing specific found:]
No existing skill/plugin directories detected.
Then ask where to save the packaged output. Build the options dynamically based on what was found:
For Claude.ai (zip):
Where should I save the zip package?
1. [Detected path, e.g., ./dist/skill-name.zip] (Recommended — existing packages here)
2. Same directory as source skill: [source-parent]/[skill-name].zip
3. Current directory: ./[skill-name].zip
4. Custom path
For Claude Code (plugin scaffold):
Where should I create the plugin scaffold?
1. [Detected path, e.g., ./plugins/skill-name-plugin/] (Recommended — existing plugins here)
2. Next to source skill: [source-parent]/[skill-name]-plugin/
3. Current directory: ./[skill-name]-plugin/
4. Custom path
Rules for suggestions:
plugins/ directory exists with other .claude-plugin/ manifests, suggest it first as recommendedskills/ directory exists, suggest the parent of that directory.zip skill packages are found, suggest the same directoryskill-name/
├── SKILL.md
├── references/
├── scripts/
└── assets/
Upload to Claude.ai:
1. Go to Settings > Capabilities > Skills
2. Click "Upload skill"
3. Select [skill-name].zip
4. Toggle the skill ON
5. Test: Ask Claude "[trigger phrase]"
Create a plugin scaffold at the location chosen in Step 3:
skill-name-plugin/
├── .claude-plugin/
│ └── plugin.json
├── skills/
│ └── skill-name/
│ ├── SKILL.md
│ ├── references/
│ ├── scripts/
│ └── assets/
└── README.md
Generate plugin.json:
{
"name": "[skill-name]",
"version": "1.0.0",
"description": "[from SKILL.md description]",
"author": { "name": "[ask user]" },
"keywords": ["[from metadata tags]"],
"license": "[from SKILL.md or MIT]"
}
Generate README.md with:
Provide installation instructions:
Install in Claude Code:
claude --plugin-dir /path/to/skill-name-plugin
Run both packaging processes. Save outputs to the locations chosen in Step 3.
Before packaging, verify:
When writing the README or marketplace description, focus on outcomes:
Good: "Set up complete project workspaces in seconds - including pages, databases, and templates - instead of spending 30 minutes on manual setup."
Bad: "A folder containing YAML frontmatter and Markdown instructions that calls MCP server tools."
If the skill enhances an MCP server, highlight both: "Our MCP server gives Claude access to your [Service]. This skill teaches Claude your team's [workflow]. Together, they enable [outcome]."
Present summary:
PACKAGED: [skill-name]
═══════════════════════
Claude.ai: [chosen-path]/[skill-name].zip (X KB)
Claude Code: [chosen-path]/[skill-name]-plugin/ (Y files)
Next steps:
- Test locally before distributing
- Run /skill-maker:test to verify trigger accuracy
- Consider hosting on GitHub for sharing
- Add to a marketplace if available
Distribution tips:
- Write README focusing on outcomes, not features
- Include example usage and screenshots
- Provide a quick-start guide (3 steps max)
- For MCP skills: document required environment and auth setup