Build production agent workflows with empirical verification - Complete teaching system with guides, diagrams, and working examples.

→ Start Here: 5-Minute Introduction | → Visual Architecture Map | → System Summary

What you'll learn: Turn "LLM analyzed my code" into "LLM + script verified with evidence" - The pattern for production-ready agent systems.

Claude Code plugins with embedded agent skills and interactive tutorials

v4.5.0: CLI 2.0 + vibe-guide plugin + Website Unification (521 routes).

Latest: v4.5.0 Release

/plugin marketplace add jeremylongshore/claude-code-plugins
/plugin install devops-automation-pack@claude-code-plugins-plus

⚡🪽 Buy me a Red Bull – Give these plugins wings and support ongoing development.

Discover, install, and manage plugins from your terminal. The CLI reads from this repo's published catalog—no manual cloning required.

pnpm add -g @intentsolutionsio/ccpi

Or with npm/yarn:

npm install -g @intentsolutionsio/ccpi
yarn global add @intentsolutionsio/ccpi

# Search plugins by keyword
ccpi search devops

# List all available plugins
ccpi list

# List installed plugins
ccpi list --installed

# Install a plugin
ccpi install ansible-playbook-creator

# Install multiple plugins at once
ccpi install devops-automation-pack security-pack api-development-pack

# Update all installed plugins
ccpi update

# Show plugin details
ccpi info project-health-auditor

# Validate plugin structure (for developers)
ccpi validate ./my-plugin

• This repo is the source of truth — All plugins live in plugins/ and catalog in .claude-plugin/marketplace.extended.json
• Website publishes versioned catalog — On merge to main, the catalog JSON is deployed to claudecodeplugins.io
• CLI consumes catalog with caching — Fetches latest catalog on first use, caches locally for speed

• Catalog updates: Happen automatically when PRs merge to main
• CLI releases: Only when CLI code changes (version bumps in packages/cli/)
• Plugin updates: Run ccpi update to pull latest versions

Huge thanks to our community contributors who make this marketplace better:

• @TomLucidor (Tom) - 🏆 TRANSPARENCY CHAMPION: His simple question "Which plugins require paid APIs vs free/self-hosted?" sparked the entire "Make All Plugins Free" initiative (v4.1.0). His follow-up about rate limits and registration requirements drove 2,400+ lines of comprehensive constraint documentation across 6 plugins (v4.3.0). Tom doesn't write code, but he keeps projects honest by asking the questions users are afraid to ask. (#148)
• @alexfazio (Alex Fazio) - 🎓 Shared production agent workflow patterns and validation techniques that inspired the complete Learning Lab system (v4.0.0). His insights on empirical verification transformed how we teach agent development.
• @lukeslp (Lucas Steuber) - 🎉 MASSIVE CONTRIBUTION: Created geepers-agents with 51 specialized agents for development workflows, deployment, quality audits, research, and game development - the largest single plugin contribution to date! (#159)
• @beepsoft - Provided valuable feedback on skill implementations that inspired comprehensive quality enhancements across the entire plugin ecosystem (#134)
• @CharlesWiltgen (Charles Wiltgen) - Created Axiom, battle-tested iOS development plugin with 13 production-ready skills for Swift/Xcode debugging and optimization (#121)
• @aledlie (Alyshia Ledlie) - Fixed 7 critical JSON syntax errors and added production CI/CD patterns from real-world debugging (#117, spotlight)
• @JackReis (Jack Reis) - Contributed neurodivergent-visual-org plugin with ADHD-friendly Mermaid diagrams and accessibility features (#106)
• @terrylica (Terry Li) - Built prettier-markdown-hook with zero-config markdown formatting and comprehensive documentation (#101)
• @clickmediapropy - Reported mobile horizontal scrolling bug, improving website UX (#120)
• @BayramAnnakov (Bayram Annakov) - 🧠 Created claude-reflect, a self-learning system that captures corrections during sessions and syncs them to CLAUDE.md - teaching Claude to remember your preferences! (#241)
• @jleonelion (James Leone) - 🔧 Fixed critical bash variable scoping bug in Learning Lab scripts and improved markdown formatting for better copy-paste experience (#239)

Want to contribute? See 000-docs/007-DR-GUID-contributing.md or reach out to [email protected]

Option 1: CLI (Recommended)

pnpm add -g @intentsolutionsio/ccpi
ccpi install devops-automation-pack

Option 2: Claude Built-in Commands

/plugin marketplace add jeremylongshore/claude-code-plugins
/plugin install devops-automation-pack@claude-code-plugins-plus

Already using an older install? Run /plugin marketplace remove claude-code-plugins and re-add with the command above to switch to the new claude-code-plugins-plus slug.

Browse the catalog: Visit Claude Code Skills & Plugins Marketplace or explore plugins/

Learn to build: See Learning Paths for step-by-step guides

739 production-ready Agent Skills (500 standalone + 239 embedded) that activate automatically based on your conversations:

Agent Skills are intelligent automation tools that Claude Code can invoke automatically when relevant. Unlike plugins that need manual commands, Skills detect when they're needed and activate seamlessly.

• 🔍 SEO Advantage: "Claude Skills" gets 3x more searches than "Claude Plugins"
• 🎯 Auto-Activation: No need to remember commands - Skills trigger automatically
• 🛡️ Tool Permissions: Fine-grained control over what each skill can do
• 📈 Future-Proof: Aligned with Anthropic's 2025 vision for AI agents

• 🔧 DevOps Skills: Automated CI/CD, deployment, monitoring
• 🔒 Security Skills: Code auditing, vulnerability scanning, compliance
• 📊 Data Skills: Analysis, visualization, reporting
• 🎨 Frontend Skills: UI/UX improvements, responsive design, accessibility
• 🧪 Testing Skills: Automated test generation, coverage analysis
• 📝 Documentation Skills: Auto-documentation, API specs, tutorials

• 739 Total Skills: 500 standalone + 239 embedded in plugins
• 20 Standalone Categories: DevOps, Security, ML, Data, API, and more
• 2025 Schema: Validated against Anthropic standards
• Average skill size: 3.2KB of intelligent automation

What are Agent Skills? They're instruction manuals that teach Claude Code when and how to use your installed plugins automatically.

1. DISCOVERY (Marketplace)
   └─ You browse claudecodeplugins.io
   └─ Find "ansible-playbook-creator"
   └─ Install: /plugin install ansible-playbook-creator@claude-code-plugins-plus

2. INSTALLATION (Files Copied)
   └─ Plugin files copied to your machine
   └─ Including skills/skill-adapter/SKILL.md ← The instruction manual!

3. STARTUP (Claude Learns)
   └─ Claude Code reads SKILL.md frontmatter from ALL installed plugins
   └─ Loads trigger phrases: "ansible playbook", "automate deployment"
   └─ Now Claude knows this plugin exists and when to use it

4. USAGE (Automatic Activation)
   └─ You: "Create an Ansible playbook for Apache"
   └─ Claude: Sees "ansible playbook" trigger → reads full SKILL.md
   └─ Claude: Activates plugin with correct workflow automatically!

WITHOUT Agent Skills:

You: "Create ansible playbook"
Claude: "I have ansible-playbook-creator installed somewhere...
         Let me manually search and figure out how to use it..."
Result: ❌ Plugin sits unused, you have to name it explicitly

WITH Agent Skills:

You: "Create ansible playbook"
Claude: *Recognizes trigger phrase instantly*
        *Reads SKILL.md for workflow*
        "I'll use ansible-playbook-creator for this!"
        *Automatically applies best practices*
Result: ✅ Instant activation, correct usage, zero thinking

Each plugin gets ONE skill file teaching Claude:

---
name: Creating Ansible Playbooks
description: |
  Automates Ansible playbook creation. Use when you need to automate
  server configurations or deployments. Trigger with "ansible playbook"
  or "create playbook for [task]".
---

## How It Works
1. Receives user request with infrastructure details
2. Generates production-ready Ansible playbook
3. Includes best practices and security configurations

## When to Use This Skill
- Automate server configuration tasks
- Deploy applications consistently
- Create repeatable infrastructure setups

## Examples
User: "Create ansible playbook to install Apache on Ubuntu"
Skill activates → Generates playbook → Ready to deploy

• ✅ Not creating new plugins - Adding instruction manuals to existing ones
• ✅ Automatic activation - Claude recognizes trigger phrases
• ✅ Best practices built-in - Each skill teaches optimal workflows
• ✅ One skill per plugin - Comprehensive instruction manual
• ✅ Only for installed plugins - Not for discovering new ones

Status: 739 skills deployed! 500 standalone skills across 20 categories + 239 embedded in plugins. All validated against 2025 schema.

Resources:

• Internal Guide: How Agent Skills Work →
• External Deep Dive: Claude Skills Technical Analysis → - Comprehensive technical breakdown by Lee-Han Chung

NEW: The first plugin using Anthropic's Agent Skills feature (launched Oct 16, 2025). Say "create a plugin" or "validate this plugin" and Claude automatically uses these model-invoked capabilities:

• 🛠️ Plugin Creator - Auto-scaffolds plugins
• ✅ Plugin Validator - Auto-validates structure
• 📦 Marketplace Manager - Auto-manages catalog
• 🔍 Plugin Auditor - Auto-audits security
• 🔢 Version Bumper - Auto-handles versions

/plugin install skills-powerkit@claude-code-plugins-plus

NEW CATEGORY: Plugins that extend Claude's built-in Skills (web_search, web_fetch) with automation. Claude searches → Plugin acts.

Example: web-to-github-issue - Research → GitHub tickets

Explore Skill Enhancers →

This marketplace contains two types of plugins that work differently:

• What they are: Markdown instructions that guide Claude's behavior
• How they work: Tell Claude HOW to perform tasks using its built-in capabilities
• Examples: DevOps pack, Security pack, API development tools
• Count: 249 plugins
• No external code execution - work entirely through Claude's interpretation

• What they are: TypeScript/JavaScript applications
• How they work: Run as separate Node.js processes that Claude can communicate with
• Examples: project-health-auditor, conversational-api-debugger, domain-memory-agent
• Count: 7 plugins (21 MCP tools total)
• Actual compiled code - 13-26KB of executable JavaScript per plugin

MCP servers load tool schemas into context. Manage active servers to optimize token usage:

/mcp                              # See active MCP servers
/mcp disable design-to-code       # Disable unused server
/mcp enable project-health-auditor # Enable when needed
/context                          # Check context usage

Full Guide: ZCF Integration & Tool Routing →

739 Agent Skills across the marketplace - automatic capabilities that Claude activates based on conversation context (500 standalone + 239 embedded in plugins).

• What they are: SKILL.md files that teach Claude when and how to use the plugin
• How they work: Claude reads trigger phrases and activates skills automatically
• Example: Say "create a plugin" and Claude uses the Skills Powerkit automatically
• Invocation: Automatic - no /command needed
• Launched: October 16, 2025 by Anthropic

Skills vs Commands: Commands require explicit /command trigger. Skills activate automatically based on what you're asking for.

Note: Plugins can bundle Skills, Commands, Agents, and MCP servers together. Agent Skills are a feature that enhances any plugin type.

Technical Analysis: How Claude Skills Work →

A comprehensive technical breakdown by Lee-Han Chung covering:

• How Skills Activate: Understanding trigger phrases and automatic invocation
• Tool Permission System: Deep dive into allowed-tools and security boundaries
• Architecture Patterns: Design principles for effective skill-based plugins
• Best Practices: Real-world examples and common pitfalls to avoid

This article is the definitive external resource for understanding how Agent Skills work under the hood.

739 skills in this marketplace are built on these principles.

# Install multiple plugins at once (CLI)
ccpi install project-health-auditor conversational-api-debugger git-commit-smart

# Or install individually (Claude built-in)
/plugin install project-health-auditor@claude-code-plugins-plus

# Then use the plugin commands
/analyze /path/to/repo    # Analyze codebase health
/debug-api                # Debug API failures
/gc                       # Smart commit messages

Share custom workflows across your organization:

// .claude/settings.json
{
  "extraKnownMarketplaces": {
    "claude-code-plugins": {
      "source": {
        "source": "github",
        "repo": "jeremylongshore/claude-code-plugins"
      }
    }
  },
  "enabledPlugins": {
    "project-health-auditor@claude-code-plugins-plus": true,
    "conversational-api-debugger@claude-code-plugins-plus": true
  }
}

Use our templates to build your own plugins:

# Copy a template
cp -r templates/command-plugin my-awesome-plugin

# Validate your plugin structure
ccpi validate ./my-awesome-plugin

# Test locally
/plugin marketplace add ./my-test-marketplace
/plugin install my-awesome-plugin@test

• Installation & Usage - Install and use plugins
• Creating Your First Plugin - Step-by-step tutorial
• Plugin Reference - Technical specifications

• Marketplace Guide - Distribute your plugins
• Security Best Practices - Secure plugin development
• MCP Server Status - MCP plugin configurations
• ZCF Integration - Tool routing, BMAD workflows, token optimization

• Production Playbooks Index - 11 comprehensive guides (~53,500 words)
• Advanced Tool Use - Dynamic discovery & programmatic orchestration ⭐ NEW
• Multi-Agent Rate Limits - Prevent API throttling
• Cost Caps & Budget Management - Control AI spending
• MCP Server Reliability - Self-healing MCP servers
• Ollama Migration Guide - Switch to self-hosted LLMs
• Incident Debugging - SEV-1/2/3/4 response protocols
• Self-Hosted Stack Setup - Full infrastructure deployment
• Compliance & Audit - SOC 2, GDPR, HIPAA, PCI DSS
• Team Presets & Workflows - Team standardization
• Cost Attribution System - Multi-dimensional cost tracking
• Progressive Enhancement - Safe feature rollout

We welcome community plugin submissions! This ecosystem thrives on shared knowledge and collaboration.

1. Fork this repository
2. Add your plugin to plugins/community/your-plugin/
3. Update .claude-plugin/marketplace.extended.json with your plugin entry
4. Run pnpm run sync-marketplace (or npm run sync-marketplace) to regenerate .claude-plugin/marketplace.json
5. Submit a pull request using our plugin submission template

See 000-docs/007-DR-GUID-contributing.md for detailed guidelines.

• Valid .claude-plugin/plugin.json
• Comprehensive README.md with examples
• 000-docs/001-BL-LICN-license.txt file (MIT or Apache-2.0 recommended)
• Tested locally and working
• No hardcoded secrets or credentials
• All scripts executable (chmod +x)

If you're actively developing your plugin and want us to stay in sync with your repo:

1. Open an issue requesting external sync
2. We add your repo to sources.yaml
3. Daily sync pulls your latest changes at midnight UTC
4. You maintain ownership - keep developing in your repo
5. Attribution preserved - .source.json tracks provenance

Benefits:

• No stale copies - your updates appear within 24 hours
• You control your codebase - we mirror it
• Full attribution - clear provenance tracking

Currently synced from:

• @numman-ali/n-skills - gastown, zai-cli

See sources.yaml for the full list of synced sources.

Claude Code plugins do NOT use GitHub Marketplace. They operate in a completely separate ecosystem using JSON-based marketplace catalogs hosted in Git repositories. This repository IS a Claude Code plugin marketplace.

There is currently no monetization mechanism for Claude Code plugins. All plugins in the ecosystem are free and open-source. See Monetization Alternatives for external revenue strategies.

Claude Code plugins are in public beta (October 2025). Features and best practices may evolve. This marketplace will stay updated with the latest changes.

Start building your own plugin today:

All templates are in the templates/ directory with complete examples.

• Claude Code Documentation
• Plugin Guide
• Plugin Reference
• Announcement Blog
• Best Practices

• Claude Developers Discord - 40,000+ members
• Report Issues
• Discussions
• Awesome Claude Code - Curated resources

• Dan Ávila's Marketplace - DevOps & productivity
• Seth Hobson's Agents - 80+ specialized subagents
• CCPlugins - Professional commands

• Plugin Packs: 4 (62 total plugin components)
• MCP Plugins: 7 (21 total MCP tools)
• Production Plugins: 2 (git-commit-smart, overnight-dev)
• AI Agency Plugins: 6 (complete business toolkit)
• Example Plugins: 3 (hello-world, formatter, security-agent)
• Templates: 4 (minimal, command, agent, full)
• Total Marketplace Plugins: 20

To be THE definitive resource for Claude Code plugins by:

1. Educating - Clear examples showing how plugins work
2. Curating - High-quality plugins you can trust
3. Connecting - Building a vibrant community of creators
4. Growing - Setting standards as the ecosystem evolves

• Quality over Quantity - Every plugin is reviewed and tested
• Learning-Focused - Understand how plugins work, don't just use them
• First-Mover - Establishing best practices for the ecosystem
• Community-Driven - Built by developers, for developers
• Actively Maintained - Updated with latest Claude Code features

MIT License - See 000-docs/001-BL-LICN-license.txt file for details.

• Anthropic - For creating Claude Code and the plugin system
• Richard Hightower - For the 100-point skill grading rubric that improved our quality standards
• Community Contributors - Everyone who submits plugins and improvements
• Early Adopters - Users who provide feedback and help us improve

Join the conversation and connect with other plugin developers and users!

• General Discussions - Community hub for all things plugins
• Announcements - Stay updated with the latest releases
• Ideas - Suggest new plugins or features
• Q&A - Get help from the community
• Show and Tell - Share what you've built!

• Discord - Claude Code Community (#claude-code channel)
• Issues - Report bugs or request features
• Pull Requests - Contribute your own plugins

• Questions? - Open a discussion
• Found a bug? - Report an issue
• Have an idea? - Share in Ideas
• Want to chat? - Join our Discord

This repository uses a structured documentation filing system for all internal project documentation. All documentation files are stored in the 000-docs/ directory using a standardized naming convention.

NNN-CC-ABCD-short-description.ext

• NNN = Zero-padded sequence number (001-999)
• CC = Two-letter category code (PP, AT, DR, RA, etc.)
• ABCD = Four-letter document type (PROD, GUID, REPT, etc.)
• short-description = 1-4 words, kebab-case

048-RA-INDX-audit-index.md                   # Audit index report
061-DR-REFF-vertex-ai-gemini-tiers.md       # Reference documentation
086-PP-PLAN-release-v1-2-0.md               # Release plan

Full specification: See the Document Filing System v3.0 in prompts-intent-solutions master-systems folder

Status: Public Beta | Version: 4.5.0 | Last Updated: December 2025 | Skills: 739