claude-plugins

A collection of plugins that extend Claude Code, the terminal-based AI coding agent by Anthropic.

Claude Code supports extensibility through Skills and Hooks, but ships without built-in tools for tracking which skills you actually use, managing notifications across multiple sessions, or validating skill quality before sharing. This repository fills those gaps:

• Usage visualization — Track skill usage with a gamified, leveling dashboard
• Session notifications — Aggregate alerts from multiple Claude Code sessions in one panel with one-click navigation
• Publish-readiness audit — Score skills against a 100-point rubric before sharing

Quick Start

# 1. Add marketplace
/plugin marketplace add JeonJe/claude-plugins

# 2. Install a plugin
/plugin install skillbook

Plugins

Skillbook — Skill Usage Dashboard

Auto-tracks Claude Code skill usage and visualizes it as Pokemon Pokedex-style collectible cards. Skills level up and gain rarity stars the more you use them.

• Hook-based auto-tracking (zero config)
• Leveling system, 5-tier rarity (Common → Legendary), 6 achievements
• Category progress bars, 7-day trend chart, workflow recommendations
• Interactive web dashboard

Screenshots

Dashboard	Skill Cards

/plugin install skillbook
/skillbook            # Open dashboard

Requires Python 3.8+, browser

---

Claude Notify — Session Notification Panel

A floating macOS notification panel built with Hammerspoon. Collects permission requests, task completions, and other Claude Code alerts in one always-on-top window. Click any notification to jump to the source terminal/tmux pane.

• Always-on-top panel + toast banner
• Type filtering (All / IN / OK)
• Project name and session tags for multi-session identification
• Dark/Light theme, adjustable opacity and font size

Screenshots

Notification Panel	Toast

cd plugins/claude-notify && bash setup.sh

Requires macOS, Hammerspoon (CLI enabled), python3

---

Skill Audit — Skill Quality Audit

Scores skills against Anthropic's skill best practices before you publish. Runs a 6-category 100-point rubric plus security anti-pattern checks, then returns an A–D grade with prioritized fixes.

• Trigger quality, body boundaries, token efficiency, and 3 more categories
• Auto-deductions for hardcoded secrets, vague descriptions, missing validation
• Fix → re-audit → verify score improvement loop

/plugin install skill-audit
/skill-audit my-skill   # Audit a single skill
/skill-audit             # Audit all project skills

No external dependencies

Contributing

PRs welcome! See each plugin's README for details.

License

MIT