Open Brain
GitHubGitHub RepoNeeds Review

anthropics/claude-plugins-official

Anthropic’s official Claude Code marketplace exposes a standard plugin package for commands, agents, skills, and MCP configuration, while leaving third-party code trust to the installer.

GitHub
Open Source Open MarkdownOpen JSON
Source Summary

The repository is Anthropic’s official Claude Code plugin directory, split into internal and third-party entries. A plugin requires **plugin.json** and may add MCP configuration, slash commands, agents, and skills; installation uses Claude Code’s plugin system.

Practical Implication

Treat its example plugin and schema as packaging references. Keep the published name as an **immutable slug**, use displayName for UI changes, and add a rename mapping when migration is unavoidable.

Agent-Ready Context
The repository is Anthropic’s official Claude Code plugin directory, split into internal and third-party entries. A plugin requires **plugin.json** and may add MCP configuration, slash commands, agents, and skills; installation uses Claude Code’s plugin system.

Treat its example plugin and schema as packaging references. Keep the published name as an **immutable slug**, use displayName for UI changes, and add a rename mapping when migration is unavoidable.

Marketplace inclusion is not a runtime safety guarantee. Anthropic says it does not control or continuously verify the MCP servers, files, or software bundled by third parties, so inspect each plugin and its source before use.
Connected Context · Feed7 Judgment

This establishes Anthropic’s authoritative Claude-specific packaging and distribution reference, including identity stability and migration rules. Against portable plugin formats and broader catalogs, it clarifies that manifest compatibility and marketplace presence are separate from lifecycle portability, provenance, permission review, and runtime safety—especially for third-party MCP-backed entries.

cursor/pluginsCursor provides a parallel client-specific marketplace and manifest model, confirming convergence around bundled skills, agents, and MCP servers while retaining separate client governance.Introducing Agent PluginsThe shared Agent Plugins format reduces repackaging of skills and MCP servers, whereas this repository defines Claude-specific distribution, naming, commands, and agent integration.anthropics/claude-plugins-communityThe community mirror adds a reviewed distribution channel, but the official directory’s warning confirms that catalog inclusion still does not guarantee ongoing third-party runtime safety.Tencent/AI-Infra-GuardIts extension and MCP scanning offers a concrete security gate for the third-party software that this marketplace explicitly says users must inspect themselves.
Context Map
toolscodingsecurity#skills#mcp#coding-agents
Uncertainty
Marketplace inclusion is not a runtime safety guarantee. Anthropic says it does not control or continuously verify the MCP servers, files, or software bundled by third parties, so inspect each plugin and its source before use.