Claude Enterprise Libraries: Plugins, Skills & Connectors¶

Claude Enterprise provides three types of extensibility mechanisms that allow organizations to expand Claude's capabilities and integrate it with external services.

Plugins¶

Plugins are managed marketplaces that organization members can browse, install, and customize. Admins control which plugins are available to users.

Plugin management¶

Plugins are added by organization admins via Settings > Plugins > Add plugins
Official plugins are curated By Anthropic & Partners and tagged as Official or GitHub
Each plugin shows its name, last updated date, and user access level
User access can be set to Available to install or restricted

Official plugins¶

Curated by Anthropic & Partners, tagged as Official or GitHub. The catalog includes 46+ plugins covering categories like security, automation, finance, design, and productivity.

Custom plugins¶

Organizations can add custom plugins beyond the official catalog using the + Add plugins button.

Plugin structure¶

A plugin is a reusable capability package that bundles together MCP connectors, skills, slash commands, and sub-agents into a single shareable unit.

my-plugin/
├── .claude-plugin/
│   └── plugin.json        # mandatory — name, description, version
├── version.json           # triggers re-sync when changed
├── .mcp.json              # MCP connector config
├── agents/
│   └── my-agent.md        # sub-agent definitions
├── commands/
│   └── my-command.md      # slash command definitions
└── skills/
    └── my-skill/
        └── SKILL.md       # skill definition

The .claude-plugin/plugin.json file is mandatory — directories without it are ignored. Changes to version.json trigger a re-sync on next launch.

Each component is optional except plugin.json:

Component	Purpose
`.mcp.json`	MCP connector configuration for external services
`agents/`	Sub-agent definitions (markdown files)
`commands/`	Custom slash commands (markdown files)
`skills/`	Reusable behaviors with optional scripts, references, and assets

Connectors¶

Connectors are MCP servers that allow Claude to access and interact with external services and data sources. They are managed under Settings > Connectors.

Connector types¶

Type	Description
Built-in	Anthropic-managed remote MCP servers (Asana, Figma, GitHub, etc.)
Custom	Your own remote MCP server, reachable over the public internet (Anthropic connects from their cloud)
Desktop extensions	Local MCP servers for on-premises resources behind a firewall — no VPN or firewall rules needed

Security¶

Connectors use per-user OAuth. Claude inherits the individual user's permissions in the connected tool — there are no shared service accounts with elevated access. On Team and Enterprise plans, an Owner must enable a connector before members can use it, and each user authenticates individually.

Adding connectors¶

Use + Add in the Connectors settings page. You can filter by Type and Categories to find the right connector.

Skills¶

Skills extend Claude's behavior with reusable capabilities. They can contain executable code and are managed under Settings > Skills.

Skill settings¶

Setting	Description
Code execution and file creation	Allows Claude to execute code and create/edit docs, spreadsheets, presentations, PDFs, and data reports. Available on web and desktop. Applies to all org members.
Skills	Master toggle for all skills org-wide, including admin-managed organization skills. Requires Code execution and file creation to be enabled.
User-created skills	Allows team members to upload or create their own skills. Disable to lock the org to approved skills only.
Skill sharing	Allows team members to share skills with each other.
Share with organization	Allows team members to share skills with the entire organization.

Security note: Skills might contain executable code. Team members should be careful when using skills from unknown sources.

Organization skills¶

Admins can manage skills that are available to everyone in the organization via Settings > Skills > Organization skills > + Add. These are centrally controlled and visible to all org members.

Skill hierarchy¶

Organization skills (admin-managed)
└── Shared skills (team member shared, if enabled)
    └── User-created skills (individual, if enabled)