Runtime Protection

Planned product

Runtime Protection will evaluate consequential actions while AI agents work. It will give IT and Security a control point before a high-risk action executes.

Intended scope

Runtime controls are expected to cover:

tool calls and commands;
file and secret access;
network destinations; and
other agent actions that cross an organization’s policy boundary.

A policy decision may allow, warn, request approval, or block an action. The evidence for the decision should remain attached to the action and outcome.

Product boundary

Runtime Protection will receive activity from hooks, gateways, proxies, extensions, pipelines, or APIs. It will share identity, policy, and evidence capabilities with the Signals Corps platform without depending on repository scanning concepts.