AI Agent Permission Scoping

Explicitly limiting which tools, data, and actions an agent can access for a given task, rather than granting broad standing access — the practical implementation of least-privilege access for agent systems.