At BoxWorks, we’re announcing Box Agent Security controls to help admins manage what agents can access and do:
- Control access: Now you can apply classification-based policies before external agents reach sensitive files.
- Set action boundaries: Protect content by using MCP Guardrails to control which tools external agents are able to work with and which actions they’re allowed to take.
- Alert on thresholds: Notify customers when certain user, app and agent activity exceeds the limits that they’ve set.
- Track agent activity [Coming soon]: Maintain long-term visibility with session governance and audit trails that show which agent acted and what it did.
What controls would help your team adopt agents with confidence?