An AI agent can write code But where should that code actually run? Think about it. An agent might need to analyze a spreadsheet, install packages, process files or generate a report. But giving it access to your entire machine just to complete a task doesn't make sense. That's what we're working on with Veto Sandbox. We're building isolated cloud workspaces where AI agents can run code, work with files and complete tasks within defined limits. For example if an agent needs to analyze a dataset and generate a chart, it should only need access to that dataset and the tools required for the job, not your personal files, credentials or unrelated projects. The idea is to give agents their own workspace, let them complete the task, retrieve the results and shut down the environment when they're done. Of course, isolation alone doesn't solve every security problem. Permissions, resource limits and network controls are equally important, and we're designing around those boundaries. Veto Sandbox is still in development and we'll be sharing demos and progress as we move forward.

Introducing Veto Sandbox. We’re working on something new under Veto Protocol. Ai agents need more than just access to tools. They also need a secure environment where they can actually execute tasks, run code and work with files. That’s what we’re building with Veto Sandbox The idea is to give AI agents their own isolated cloud workspace where they can work within clearly defined limits. Here’s how it will work: • Create a dedicated workspace for your agent. • Upload files, run code and build applications. • Retrieve the results once the task is done. • Shut down the environment when it’s no longer needed. With $VETO we started by controlling what AI agents can access. Now with Sandbox, we’re taking that a step further by controlling where and how they execute tasks. We’re also planning APIs, SDKs, resource limits and network controls to make it easier for developers to integrate. Veto Sandbox is still in development: We’ll be sharing progress, demos and early testing opportunities as we move forward. Our goal remains the same: make AI agents more useful without compromising user control. A workspace for Ai agents. Boundaries by design.

Approval Routing is now live! When an agent needs your approval, the request shouldn’t depend on you watching a dashboard. In this walkthrough: We show the webhook configuration behind Veto’s approval alerts: connecting an HTTPS endpoint, choosing which events to receive and limiting notifications to your own events when needed. That includes both requests awaiting approval and updates when an approval has been resolved. For developers: This provides a way to connect Veto’s approval events to external notification workflows. The alert carries the update not the authority to approve an action. That decision remains within Veto’s approval process. One important detail: the version shown uses “Deliver now” to dispatch queued events. Automatic delivery and native Slack, email or push notifications are not demonstrated in this recording. The wider video also covers Policy Templates, Slack integration, Agent Fingerprints and Shadow Agents. Watch the Alerts section toward the end for the Approval Routing configuration. dApp: app.vetoprotocol.live Your agent asks. You decide.