aws2 Documentation

aws2 helps teams review AI agents as part of a real business workspace: what they can reach, what they can do, where approval is required, what evidence remains, and what the organization should not claim yet.

Why this exists

AI agents are no longer only chat windows. In many workflows they can read files, call tools, query systems, run commands, draft messages, update records, or trigger connected services. That makes the review unit bigger than a model answer and smaller than the entire company.

aws2, the Agentic Workspace Security Standard, is a profile model for one scoped agentic workspace system. The goal is to make a workflow visible enough to review: boundary, authority, action gates, source trust, context handling, sensitive data, receipts, validation, exceptions, and claim limits.

Entry paths

• Start with the plain-language primer if you are deciding whether this problem applies to your business.
• Use the applied example if you want to see what a scoped profile looks like in practice.
• Read the standard draft when you need the formal structure: scope, terms, levels, control families, evidence model, governance, and claim language.
• Use the family guides when you need more context for a specific AWS2-* family without starting from formal requirement text.