CrewAI
Multi-agent automation framework & studio to build and run AI crews
Claude Code Plugins
Modular AI agents and plugins for end-to-end software development
A production-ready marketplace of 63 focused plugins, 85 specialized agents, and 47 skills that enable token-efficient, multi-agent orchestration for full-stack development, security, DevOps, and AI workflows.
Overview
Overview
Claude Code Plugins delivers a production-ready marketplace of 63 single-purpose plugins, each bundling specialized agents and skills. With 85 domain-expert agents covering architecture, languages, infrastructure, security, data/AI, and business, plus 47 progressive-disclosure skills, the system provides token-efficient AI assistance for every stage of software creation.
Deployment & Usage
Add the marketplace with a single CLI command, then install only the plugins you need—e.g., python-development, kubernetes-operations, or full-stack-orchestration. Each installation loads its agents, commands, and skills into Claude’s context, keeping token usage low while allowing you to compose multiple plugins into complex multi-agent workflows. Hybrid orchestration automatically routes planning tasks to Sonnet and execution steps to Haiku, balancing performance and cost. The result is an extensible AI-driven development environment that scales from quick code snippets to end-to-end production pipelines.
Highlights
Granular plugin architecture with minimal token load
85 domain-expert agents covering development, ops, security, AI, and business
47 progressive-disclosure skills for efficient knowledge activation
15 multi-agent workflow orchestrators for complex end-to-end pipelines
Pros
- Token-efficient execution keeps model usage low
- Highly modular – install only needed plugins
- Broad coverage across development, operations, security, and AI
- Built-in multi-agent orchestration reduces manual scripting
Considerations
- Requires familiarity with Claude Code command syntax
- Large catalog may overwhelm new users
- Hybrid model orchestration depends on specific Anthropic models availability
- Interaction is CLI/command based without a graphical UI
Managed products teams compare with
When teams consider Claude Code Plugins, these hosted platforms usually appear on the same shortlist.
Looking for a hosted option? These are the services engineering teams benchmark against before choosing open source.
Fit guide
Great for
- Teams seeking AI-augmented full-stack development pipelines
- Organizations needing automated security scanning and code review
- Developers wanting plug-and-play AI agents for specific languages or frameworks
- DevOps engineers looking to generate infrastructure code via AI
Not ideal when
- Projects that cannot integrate the Claude Code environment
- Users preferring a single monolithic AI tool over modular plugins
- Environments with strict offline requirements (requires model access)
- Teams without expertise in prompt engineering or agent orchestration
How teams use it
Full-stack feature implementation
Automatically scaffolds backend, frontend, tests, security checks, and deployment for a new OAuth2 authentication flow.
Kubernetes production deployment
Generates Helm charts, GitOps manifests, and security policies, then applies them to a cluster with minimal manual effort.
Comprehensive security hardening
Runs multi-agent SAST, dependency scanning, and code-review audits, delivering a prioritized remediation report.
Python microservice scaffolding
Creates a FastAPI project with async patterns, pytest suite, and UV package manager configuration.
Tech snapshot
Python66%
Shell34%
Frequently asked questions
How do I install only the plugins I need?
Use `/plugin marketplace add <repo>` then `/plugin install <plugin-name>`; each installation loads only its agents, commands, and skills.
What models power the agents?
The system uses a hybrid orchestration of Haiku for fast deterministic tasks and Sonnet for complex reasoning, automatically assigned per workflow.
Can I combine multiple plugins in a single workflow?
Yes, plugins are composable; you can invoke multi-agent orchestrators that coordinate agents from different plugins.
Is there a limit to the number of agents I can run concurrently?
Token limits are managed per request; loading many agents increases token usage, so you should install only required plugins to stay within model limits.
Where can I find documentation for each plugin?
The repository includes a Plugin Reference catalog, Agent Reference, and detailed usage guides accessible via the `/plugin` command or the online docs.
Project at a glance
Active- Stars
- 26,086
- Watchers
- 26,086
- Forks
- 2,877
LicenseMIT
Repo age6 months old
Last commit2 days ago
Primary languageC#
Last synced yesterday