Claude Cowork Just Became 10x Better (Plugins Guide)

Ben AIAbout 5 min readFeb 14, 2026Watch original
THE SUMMARYAI-generated

Cloud Co-work Plugins: A Deep Dive into the Future of Work

Key Concepts:

  • Plugins: Specialized modules for Cloud Co-work, containing skills, tools, and commands to enhance AI agent capabilities. Analogized to a USB drive providing specific expertise.
  • Skills: Specific capabilities with instructions (skill MD file – essentially a system prompt) for executing tasks, potentially including knowledge sources and tool usage.
  • Connections: Cloud Co-work’s ability to integrate with software via built-in connectors, MCPs (Multi-purpose Connector Plugins), and browser access.
  • Commands (Slash Commands): Triggers to initiate specific skills or agentic workflows.
  • Agentic Workflows: Chained instructions utilizing multiple skills to automate complex tasks.
  • Skill MD File: A markdown file containing the instructions, knowledge sources, and tool usage details for a specific skill.
  • MCP (Multi-purpose Connector Plugin): A connector allowing Cloud Co-work to access and interact with various software applications.

I. Introduction: The Disruption of Plugins & SAS Market Impact

The launch of plugins for Cloud Co-work by Entropic has triggered significant market reaction, resulting in billions of dollars in wiped-out SAS (Software as a Service) market capitalization. This stems from the potential for plugins to fundamentally alter how individuals and companies approach work. The core premise is granting Cloud Co-work access to computer files and software, enabling automation through “skills.” A full tutorial on Cloud Co-work is available (link in description). Plugins are accessible via the bottom left corner of the Cloud Co-work interface.

II. Plugin Structure & Categorization

Entropic has initially categorized plugins by business department (Sales, Productivity, Product Management, Legal, Finance, Customer Support, etc.). The intention is to create specialized “experts” for each function. For example, the Customer Support plugin includes connectors to customer service platforms and skills related to customer research, response drafting, and escalation. Plugins can be invoked directly using a slash command (e.g., /go to plugins) or automatically triggered by user intent (e.g., stating “prepare for a sales call” will automatically activate the sales call prep skill).

III. Skills, Connections, and Commands: The Building Blocks

The video breaks down the core components of the plugin ecosystem:

  • Skills: Defined as specific capabilities with detailed instructions outlined in a “skill MD” file (a markdown-formatted system prompt). Skills can incorporate knowledge sources, tool usage instructions, and even sub-agent activation. An example given is a personalized LinkedIn writer skill, utilizing a skill MD file and supporting documents like profile background, ICP, and voice personality guidelines.
  • Connections: Cloud Co-work’s ability to connect to software is crucial. This is achieved through built-in connectors, MCPs, and a browser-based MCP for accessing software lacking direct integration. The speaker notes a potential security benefit of packaging connectors within plugins, limiting access based on departmental needs (e.g., preventing customer service from accessing sales software).
  • Commands: These are triggers (slash commands) that initiate specific skills. More powerfully, commands can orchestrate “agentic workflows” – sequences of skills working together to automate complex tasks. The speaker’s YouTube repurposing workflow exemplifies this, chaining LinkedIn writer, infographic creator, and newsletter writer skills with a single “repurpose” command. This is comparable to automation platforms like Zapier or Make.com.

IV. The Potential to Reshape the SAS Landscape

The introduction of plugins poses a potential threat to traditional SAS companies. The ability to encapsulate common departmental workflows within plugins, accessible to anyone without technical expertise, democratizes automation. Key arguments supporting this claim:

  • Accessibility: Plugins empower non-technical users to automate tasks and workflows across software platforms.
  • Ease of Modification: Workflows can be easily created, improved, and shared through prompting, eliminating the need for extensive coding.
  • Increased Efficiency: New employees can leverage existing plugins to perform tasks faster, benefiting from embedded domain expertise and knowledge sources.
  • Potential for AI Agent as Central Interface: Plugins could position Cloud Co-work (or similar AI agents) as the primary interface for work, replacing the need to navigate multiple software applications.

The speaker cites the stock drops of Salesforce, ServiceNow, and Adobe following Entropic’s plugin launch as evidence of this perceived threat. However, a potential counter-response from SAS companies is to develop their own plugins, integrating their unique workflows and functions into the plugin ecosystem.

V. Plugin Types & Future Outlook

The speaker anticipates three types of plugins emerging:

  1. Entropic-Built Plugins: Currently focused on common business departments, with plans for organization-wide sharing and private marketplaces.
  2. Third-Party Provider Plugins: Potentially developed by SAS companies or independent builders, offering monetizable solutions.
  3. Custom-Built Plugins: Tailored to the unique workflows and processes of individual businesses. The speaker emphasizes the importance of building custom plugins internally.

VI. Setting Up & Customizing Plugins

Two primary approaches to utilizing plugins are outlined:

  1. Customizing Existing Plugins: Entropic allows users to adapt pre-built skills and plugins through prompting (e.g., “adjust the draft outreach skill and add examples of my templates”). Full plugin customization is also available.
  2. Building Plugins from Scratch: The recommended approach involves first creating individual skills, then bundling them into a plugin, and finally chaining skills together into agentic commands.

Skills can be created by prompting Cloud Co-work with a task description and supporting documentation, or by converting existing projects/custom GPTs into skill MD files. Plugins can be created by instructing Cloud Co-work to combine specific skills, connectors, and commands.

VII. Sharing & Deployment

Currently, plugins created within a Cloud Co-work account are locally saved. Sharing requires exporting a zip file, which can then be imported into other accounts. The speaker anticipates Entropic simplifying internal and external plugin sharing in the future, potentially through marketplaces or GitHub integration. A full guide on deployment is available (link in description).

VIII. Community & Resources

The speaker promotes their AI accelerator community (link in description) as a resource for plugins, skills, weekly workshops, one-on-one tech support, and blueprints for building an AI business.

Notable Quote:

“Plugins could potentially create a whole new economy and change the SAS landscape forever.” – Speaker, emphasizing the transformative potential of the technology.

Conclusion:

Cloud Co-work plugins represent a significant advancement in AI-powered automation. By democratizing access to complex workflows and enabling customization, they have the potential to reshape how work is done, challenging the dominance of traditional SAS solutions. The key takeaway is to begin exploring plugin creation and customization to unlock the full potential of AI within your organization. The future of work may well be defined by the ability to effectively leverage these powerful tools.

AI summaries can miss context or contain errors. Check important details against the original video.

Go a little deeper.

Have a question about this video? Load its transcript to open the video chat.