Define intended agent behavior with the Agent Manifest

The Agent Manifest gives allows you to define an AI agent’s boundaries, including its purpose, expected behavior, and scope. Atlas can use this context to more accurately evaluate whether an agent’s reasoning, tool use, and data access align with its intended function.
This extends Agent IBAC’s runtime controls with a clearer baseline for determining when agents drift beyond their intended purpose, so out-of-scope actions can be stopped before they execute.




.webp)


