Claude plugins get a portal as MCP usage climbs 110x

Anthropic now has a portal for third-party plugins. Developers can submit a plugin there, follow its review and see usage numbers once it goes live, the ClaudeDevs account said on X on September 25. In the same post Anthropic called plugins the main way outside developers build extensions for Claude. Over the coming weeks, plugins will also get one discovery experience across both Claude and Claude Code.

Testing Catalog reported that the submission flow lives on Claude Directory and that it takes MCP Connectors as well as plugins. The outlet described plugins as the new “apps”.

Anthropic backed the push with one figure: MCP usage across Claude products is up 110x this year. It didn’t say what the starting level was, so the multiple can’t be turned into an absolute number.

A plugin doesn’t add capabilities of its own. It packages MCP servers and skills, and Claude Code also lets it carry hooks and subagents, so they install with one command and update from a marketplace. According to Anthropic’s Claude Code docs, each of those pieces also works without a plugin, and a plugin makes sense when you want to ship several of them together. In practice a plugin is a directory with a manifest at .claude-plugin/plugin.json. The manifest gives the plugin its name and can also hold a version, a description and other metadata.

Plugin authors already have a way to test their work before it reaches reviewers: in September Claude Code gained claude plugin eval, which runs a plugin’s eval suite and returns scored results.

Installed plugins also have a cost for users. An enabled plugin is loaded into every session. The name and description of every skill, agent and command that Claude can call on its own sit in context on every turn, and those tokens count toward usage. The docs show that cost only for plugins in Anthropic’s official marketplace: open /plugin, go to the Marketplaces tab and look for the Context cost estimate.