Claude Code Plugins: The Official Anthropic Library and How to Install Any Plugin

Anthropic just open-sourced 11 production plugins under anthropics/knowledge-work-plugins. The repo is at 17.3K stars and climbing.

If you've been looking for Claude Code plugins that cover real knowledge work — sales, legal, finance, data, product management — this is the official starting point. Here's what's inside, how to install them, and what the architecture tells you about where Claude-native work is heading.

What Are Claude Code Plugins

Claude Code plugins extend what Claude understands and can do within a session. They add domain-specific skills, slash commands, and tool connections on top of the base model.

A plugin is not a separate app or an API wrapper. It's a structured directory of:

• Skills — markdown files encoding domain knowledge, workflows, and terminology Claude draws on automatically
• Commands — explicit slash triggers (e.g. /sales:call-prep, /data:write-query)
• Connectors — MCP servers wiring Claude to external tools (Slack, Notion, HubSpot, Snowflake, etc.)

The key insight from the official Anthropic library: every plugin is markdown and JSON. No code. No build steps. Any knowledge worker can open a skill file, read it in plain English, and customize it for their situation.

The 11 Official Plugins

productivity — tasks, calendars, personal context. Connects to Slack, Notion, Asana, Linear, Jira, Monday, ClickUp, Microsoft 365.

sales — prospect research, call prep, pipeline review, outreach drafts, battlecards. Connects to HubSpot, Close, Clay, ZoomInfo, Fireflies.

customer-support — ticket triage, draft responses, escalation packaging, knowledge base creation. Connects to Intercom, HubSpot, Guru.

product-management — specs, roadmaps, user research synthesis, stakeholder updates, competitive tracking. Connects to Linear, Figma, Amplitude, Pendo, Intercom.

marketing — content drafts, campaign planning, brand voice enforcement, competitive briefs, performance reporting. Connects to Canva, Figma, HubSpot, Ahrefs, Klaviyo.

legal — contract review, NDA triage, compliance, risk assessment, templated responses. Connects to Box, Egnyte, Jira.

finance — journal entries, reconciliation, financial statements, variance analysis, close management, audit support. Connects to Snowflake, Databricks, BigQuery.

data — SQL, statistical analysis, dashboards, validation. Connects to Snowflake, Databricks, BigQuery, Hex, Amplitude.

enterprise-search — one query across email, chat, docs, wikis. Connects to Slack, Notion, Guru, Jira, Asana, Microsoft 365.

bio-research — preclinical research tools, literature search, genomics, target prioritization. Connects to PubMed, BioRender, ChEMBL, ClinicalTrials.gov.

cowork-plugin-management — create and customize plugins for your organization.

Plugin Structure

Every official plugin follows the same layout:

plugin-name/
├── .claude-plugin/plugin.json   # Manifest
├── .mcp.json                    # Tool connections
├── commands/                    # Slash commands you invoke explicitly
└── skills/                      # Domain knowledge Claude uses automatically

Skills are markdown files. Commands are explicit triggers. Connectors wire Claude to external tools via MCP servers.

Anthropic's README is explicit: these are generic starting points. They get useful when you customize the skill files for how your company actually works — your terminology, your process, your specific tools.

How to Install Claude Code Plugins

If you're on Claude Cowork: Install directly from claude.com/plugins.

If you're on Claude Code:

claude plugin marketplace add anthropics/knowledge-work-plugins
claude plugin install sales@knowledge-work-plugins

Once installed, skills activate automatically when relevant. Slash commands become available in your session — e.g. /sales:call-prep, /data:write-query, /marketing:draft-content.

To customize for your company: Fork the repo. Edit the markdown skill files — add your company's terminology, internal process steps, and specific tools. Commit to your org's repo. Point Claude at your fork.

This is intentional: the file-based design means customization requires no engineering. A sales manager can open the sales/skills/ directory, read what's there in plain English, and add the language their team actually uses.

What It Signals About Claude-Native Work

Anthropic is building Claude Cowork as a knowledge work platform, not just a coding tool with enterprise features bolted on. The knowledge-work-plugins repo is the reference implementation of how they think that platform should be configured.

The bet: AI-native work isn't about replacing knowledge workers. It's about giving them a Claude that understands their specific role, tools, terminology, and processes.

The mechanism: plugins as institutional knowledge. Your org's legal standards live in a skill file. Your sales process lives in a command. Your finance workflow lives in a connector. Version-controlled, shareable, improveable.

For builders, this is also a template for product design. The patterns here — skills plus commands plus connectors, all file-based — are the same patterns you'd use to build a specialized product on top of Claude. The difference between an internal company plugin and a commercial product is marketing and distribution, not the underlying architecture.

Bottom Line

If you're looking for Claude Code plugins that cover real knowledge work verticals, anthropics/knowledge-work-plugins is the place to start. The official library covers 11 roles, all customizable via markdown files, all wired to the tools those roles actually use.

Install the relevant plugin, read the skill files, customize for your context, and you have a Claude that knows your domain — not just a generic LLM you're prompting from scratch every session.

AI Builder Club members build this kind of AI-native tooling every week. Start with our free Claude Code 101 course to master the foundation, then join the community for weekly workshops and premium boilerplates.