- Home
- Skills
- Agents & Orchestration
- Claude Plugin Packager: Ship Skills as a Plugin
Claude Plugin Packager: Ship Skills as a Plugin
Manual plugin packaging for Claude Code is tedious and prone to path errors.
Free
Works with the AI Agents you already use
Claude Plugin Packager: Ship Skills as a Plugin
Example session with this skill installed
Package my custom 'git-helper' skill and 'deploy' slash command into a plugin called 'dev-tools' version 1.1.0 with an MIT license.
- Read your context and instructions
- Compiled the claude plugin packager
I have inventoried your assets and prepared the 'dev-tools' plugin package. I have normalized all paths and generated the required manifest files.
Plugin Structure
dev-tools/
├── .claude-plugin/
│ └── plugin.json
├── skills/
│ └── git-helper/
│ └── SKILL.md
├── commands/
│ └── deploy.md
├── README.md
├── CHANGELOG.md
└── LICENSE
Generated plugin.json
{
"name": "dev-tools",
"version": "1.1.0",
"description": "Streamlined git workflows and deployment automation for Claude Code",
"author": {
"name": "Developer"
},
"license": "MIT",
"keywords": [
"git",
"deployment",
"automation",
"workflow"
]
}
Installation Instructions
To test this plugin locally, run the following commands in your terminal:
-
Add your local directory as a marketplace:
/plugin marketplace add ./dev-tools -
Install the plugin:
/plugin install dev-tools -
Verify the installation:
/plugin list
Next steps
- Review the
README.mdto ensure the description matches your expected usage. - Run
/deployto verify the slash command is correctly registered. - Commit the
dev-toolsfolder to your repository to share it with your team.
Connects securely to your tools. The creator never sees your data.
About this skill
Manual plugin packaging for Claude Code is tedious and prone to path errors. This skill automates the transition from local scripts and markdown files to a fully structured, versioned, and installable plugin or marketplace.
What it does
- Inventory discovery: Scans
.claudedirectories to find existing skills, slash commands, agents, and hooks. - Manifest generation: Creates valid
plugin.jsonandmarketplace.jsonfiles with correct semver and metadata. - Path normalization: Replaces absolute paths in MCP configurations and hooks with the
${CLAUDE_PLUGIN_ROOT}variable. - Quality validation: Checks for missing frontmatter, hardcoded secrets, and executable permissions on scripts.
- Distribution setup: Generates a standardized directory tree, README, and installation instructions for GitHub.
How it works
- Analyze: The skill inventories your local
.claudeassets and asks which items to include in the bundle. - Structure: It creates the plugin directory layout, moving skills, commands, and agents into their respective folders.
- Configure: It generates the
plugin.jsonmanifest and, if requested, themarketplace.jsonfor multi-plugin repos. - Validate: It runs a quality gate to ensure all hooks are quoted and skills have trigger-rich descriptions.
- Finalize: It provides the tree structure and the exact Claude Code commands needed for a test install.
Frameworks & tools
Specifically designed for Claude Code CLI and the .claude configuration ecosystem. Works with MCP servers defined in .mcp.json.
Why this beats prompting it yourself
Generic prompts often fail to use the ${CLAUDE_PLUGIN_ROOT} env variable correctly, leading to broken hooks after installation. This skill enforces the exact folder hierarchy and manifest schema required by the Claude Code plugin architecture.
Use cases
- Convert a collection of local
.mdskills into a sharable plugin. - Create a private marketplace for team-wide agent and command distribution.
- Package MCP servers with pre-configured hooks for automated workflows.
- Standardize versioning and changelogs for internal Claude Code tools.
Known limitations
Does not handle binary compilation of MCP servers. Scripts must be pre-built or interpreted. Requires existing .claude folder assets as input.
Does not compile binary MCP servers. Requires pre-existing .claude assets. Assumes usage within the Claude Code CLI environment.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
This skill has been used 1 time.
Trust & safety
Security scan
Security scan passed 4 days ago
- Free with an Agensi account
Permissions
No special permissions declared or detected