Google Cloud has opened Auth Manager to general use, giving AI agents a managed store for delegated credentials and automatic token renewal. A separate preview adds identity-based authentication across cloud and internal services. The changes accompany code-security updates, including isolated vulnerability testing and human review of actions such as writing files or running commands.
Google Cloud, Google’s cloud-services division, opened Auth Manager to general use on 5 November. The service stores credentials delegated to AI agents in a vault managed by Google. It also handles token renewal. The change concerns the software’s access to services while performing tasks for users. Google presents the managed credential store as part of its controls for agent authentication.[1], [2]
Agent Identity support for JSON web tokens (JWT) remains in preview. Google describes authentication across Amazon Web Services, Microsoft Azure, internal systems and Model Context Protocol servers, which connect models to tools and data. The company presents the approach as a way to reduce reliance on static, long-lived application programming interface keys. General availability for Auth Manager and preview status for this identity support are separate.[1]
The same announcement updates CodeMender, Google’s code-security tool in preview since July. Vulnerabilities can be validated in an operating-system sandbox or an isolated virtual machine. CodeMender scans can cover an entire repository, a chosen folder or modified files. Local writes and command execution are subject to human review. The workflow separates identifying a defect from testing a proposed correction before changes are applied.[1]
