Author(s): Rick Hightower

Originally published on Towards AI.

Claude Code Plugins 14: Plugins are how to package the skills, subagents, hooks, and MCP servers you have already built into a Claude Code plugin your whole team installs in a single command.

You did the work. You wrote a /ship skill that runs your project’s verification chain. You built a code-reviewer subagent that catches what the first-draft writer misses. You wired up a PostToolUse hook that formats every file on save, and an MCP server that lets Claude check migrations against the live database. Every one of those pieces earns its keep.

After explaining that Claude Code customizations are stranded in a single repo’s .claude/ directory, the article defines what a plugin really is (a directory-shaped bundle of skills, agents, hooks, MCP servers, and other components managed by a .claude-plugin/plugin.json manifest). It covers two key behaviors—namespacing to avoid collisions and the fact that plugins are copied (so external paths break unless you use ${CLAUDE_PLUGIN_ROOT} at runtime). The post then shows how to install plugins from the official marketplace, outlines categories like LSP/code intelligence and common integrations, and mentions marketplace features like context-cost estimates. It goes on to describe additional marketplaces (community and a demo marketplace), installation scopes that determine whether plugins apply to user, project, or local settings, and the commands and reload flow needed to activate them. Next, it walks through when it’s worth building your own plugin, the local “quickstart” structure to begin iterating, how to add skills (including how they’re invoked), and how to add hooks and MCP servers safely—especially around using ${CLAUDE_PLUGIN_ROOT} for bundled files and ${CLAUDE_PLUGIN_DATA} for persistent state. The article closes with validation/testing commands, common path-related failure modes, and practical distribution strategies for personal use, team use (committing a project-scoped plugin and enabling it via project settings.json), and public distribution via a marketplace marketplace.json plus submission and safety checks.

Read the full blog for free on Medium.

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI