A curated collection of plugins, rules, agents, and commands for Claude Code.
This repo is a plugin marketplace for Claude Code — a central place for reusable plugins, organization-wide rules, custom agents, and slash commands. It serves two purposes:
- Plugin distribution: 27 plugins covering git automation, safety evaluation, status lines, task management, and more
- Organization-wide AI configuration: Rules, agents, and commands that get synced to
~/.claude/for consistent behavior across all projects
Add the plugin path to your project's .claude.json:
{
"plugins": ["/path/to/this/repo/plugins/scm-utils", "/path/to/this/repo/plugins/statusline"]
}The .ai/rules/ directory is synced to ~/.claude/rules/ via automation (symlinks). This makes rules available across all projects without per-project configuration.
| Plugin | Description |
|---|---|
| commit-command | /commit — AI-generated commit messages matching your repo's style |
| commit-skill | Auto-analyze changes and create semantic commits during development |
| scm-utils | /commit, /update-branch commands + auth-user skill |
| git-spice | Manage stacked Git branches with the gs CLI |
| Plugin | Description |
|---|---|
| safety-evaluation-prompt | Pre-tool-call safety via prompt-style hooks |
| safety-evaluation-script | Pre-tool-call safety via script-style hooks (haiku CLI) |
| context-bloat-prevention | PostToolUse + PreToolUse hooks to detect and prevent context bloat |
| Plugin | Description |
|---|---|
| statusline | Configurable status line showing session info, project context, git status |
| statusline-iterm | Status line with iTerm2 badge integration |
| Plugin | Description |
|---|---|
| review-changes | /review-changes — detailed code review feedback |
| code-simplifier | /simplify — refine code for clarity and maintainability |
| create-command | /create-command — guided slash command creation |
| correct-behavior | /correct-behavior — fix AI behavior mistakes and update rules |
| product-development-and-sdlc | Iterative PRD writing with structured SDLC workflows |
| Plugin | Description |
|---|---|
| task-parallelization | Intelligently parallelize Task tool calls for batch operations |
| todo-sync | Auto-sync ~/.claude/ todos to project .claude/ directory |
| self-terminate | Graceful SIGINT termination for agents |
| Plugin | Description |
|---|---|
| tmux-subagent | /subagent — launch sub-agents in tmux with custom configs |
| agent-teams-skills | Reference skill for agent teams: enabling, config, hooks, tmux |
| Plugin | Description |
|---|---|
| linear-mcp-sync | Linear MCP with hash validation hooks for safe ticket updates |
| github-auth-skill | GitHub device authorization flow authentication |
| sync-settings | Sync local Claude Code settings via syncconfig.yaml rules |
| Plugin | Description |
|---|---|
| data-serialization | YAML/JSON/TOON/XML conversion; TOON reduces tokens 30-60% |
| memory-manager | Auto-detect and store user preferences in CLAUDE.md |
| command-help-skill | Help discover and execute slash commands sent as messages |
| skills-maintenance | Maintain, update, and improve existing Claude Code skills |
| opengraph-image | Generate OpenGraph images via html2png.dev API |
21 behavioral rules in .ai/rules/ covering:
- Code quality: DRY, KISS, YAGNI, incremental development
- Git workflow: PR conventions, commit hygiene, branch management
- Task management: Todo tracking, task planning, sub-agent delegation
- Communication: Speech-to-text handling, intellectual honesty, polite correction
- Safety: Verify before blaming, never say done prematurely, error handling
Rules are symlinked to ~/.claude/rules/ by automation and apply across all projects.
6 custom agents in .ai/agents/:
| Agent | Purpose |
|---|---|
conversation-history-search |
Search past Claude Code conversations (haiku model) |
github-issue-creator |
Create GitHub issues for bugs and work items |
internet-researcher |
Deep web research with reference gathering |
research-lead |
Lead and coordinate multi-source research |
research-subagent |
Focused research sub-agent for deep dives |
ui-ux-consultant |
Desktop UI/UX and accessibility expert |
4 commands in .ai/commands/:
| Command | Purpose |
|---|---|
/correct-behavior |
Fix AI behavior mistakes and update rules |
/create-command |
Guided slash command creation |
/review-changes |
Detailed code change review |
/relentlessly-fix |
Persistent fixing until resolved |
mise run lint # Run all linters
mise run validate # Validate plugin structure
mise run check # Run lint + validateTo add a plugin:
- Create a directory in
plugins/your-plugin-name/ - Add
.claude-plugin/plugin.jsonwith name, version, description - Add your commands, skills, hooks, or agents
- Submit a pull request
plugins/your-plugin-name/
├── .claude-plugin/
│ └── plugin.json # Required: name, version, description
├── commands/ # Slash commands (*.md)
├── skills/ # Agent skills (*/SKILL.md)
├── hooks/ # Lifecycle hooks
├── agents/ # Agent definitions
└── README.md # Optional
- claude-team — CLI tool for launching agent team sessions
- claude-utils — CLI helper scripts for Claude Code (Homebrew)
- agent-team — POC for provider-agnostic agent orchestration
Proprietary. All rights reserved.