Shield Agent AI
Use cases

Same rules on different jobs

Each case below uses identity, policy, human approval, and a trail. The agent must call evaluate. Shield does not find it on the network. How you use it.

Platform and ops agents

Hold deploys, restarts, and external API side effects until someone approves. Kill switch and event log included.

Compliance and labeling agents

Agents that touch regulated work use the same approval and trail rules as everything else.

Coding-agent gates

Watch IDE or CLI sessions, raise risk on unknown tools, and block irreversible actions from running alone.

Any irreversible side effect

Writes, exports, refunds, network submits. Same path: policy, human approval when required, then trail.

Install How KYA works