Everything you need to know about Claude Cowork — from what it is to how it runs, in one shareable reference.
Cowork uses the same agentic architecture that powers Claude Code, now accessible within Claude Desktop. Instead of responding one message at a time, Claude can take on complex, multi-step tasks and execute them on your behalf.
Cowork has a few prerequisites and a simple 3-step launch. Here's everything to know before you begin.
Cowork runs directly on your computer, giving Claude access to the files you choose to share. Code runs safely in an isolated virtual machine.
Three power features that let you customize, automate, and specialize your Cowork experience.
A Project is a dedicated workspace with its own files, context, instructions, and memory — backed by a real folder on your computer.
Plugins bundle skills, connectors, and sub-agents into a single package. Install one and Claude becomes an expert for your role. 15 open-source plugins available to customize and build upon.
Set up tasks that Claude runs automatically or on-demand — something that isn't possible in regular chats outside of Cowork.
Work that fits Cowork's shape — volume, parallelism, in-place results. These are the tasks where Cowork shines.
Cowork is built with safety by design. Here's what you need to know about how it protects your files and manages usage.
Cowork stores conversation history locally on your computer. Activity is not captured in Audit Logs, Compliance API, or Data Exports. Do not use Cowork for regulated workloads.
Cowork tasks consume more usage than standard chats because multi-step tasks are compute-intensive. Here's how to get the most from your allocation.
Pick a plugin that matches your role. Install it in seconds, then hand Cowork a real task from your week.
Introduction to Claude Cowork — Reference Cheatsheet