Anti-pattern imho. Agents should operate within granular identity and permission scopes, with audit and log trails for all data operations (read, write, etc).

Anti-pattern imho. Agents should operate within granular identity and permission scopes, with audit and log trails for all data operations (read, write, etc). Copilot: https://learn.microsoft.com/en-us/entra/agent-id/identity-platform/what-are-agent-identities https://learn.microsoft.com/en-us/entra/agent-id/identity-pl... | https://learn.microsoft.com/en-us/purview/audit-copilot https://learn.microsoft.com/en-us/purview/audit-copilot (for example) TLDR Maintain an identity boundary whenever possible.