How do I set up a Claude Project for my project?

A Claude Project is a workspace that keeps reference files and standing instructions together, so every new conversation starts with your project's context. In the book, this is where your Project Brain lives. Chapter 1, The Project's Brain, and Chapter 2, Feeding the Brain, walk through it in detail.

Menus, file limits, and which plans include Projects change over time, so check Anthropic's current help pages for the exact steps.

Set it up

1. Create one workspace per project. Name it clearly, for example "Website Relaunch 2026". Keeping projects separate stops details from one leaking into another.
2. Write the standing instructions. Tell the assistant who you are, what the project is, who reads your outputs, and how you want answers formatted. Keep it short and specific.
3. Upload your core documents. Start with the charter, scope, schedule, stakeholder list, and a glossary of acronyms. See What documents should I give my AI assistant? for the full list.
4. Test it. Open a new chat and ask something only your documents can answer, such as "Who approves scope changes?" If the answer is vague or wrong, fix the source document.

Keep it working

• Replace, don't pile up. When a document changes, remove the old version so the assistant doesn't read two conflicting copies.
• Add recent news regularly. Weekly notes and decisions keep the Brain current.
• Clean up on a schedule. The book calls this the Janitor Protocol. Read What is the Janitor Protocol?.

Before uploading anything, check your organisation's AI policy and the provider's data settings. See Is it safe to share project data with AI tools?.

For a walkthrough, read How to Build Your Own AI-Powered PMO with Claude and Stop Repeating Your Project Context to AI Every Single Time.