· ai-devtools
Anthropic releases self-hosted Claude Managed Agents for on-premise credential storage
Agent credentials can now reside within your own infrastructure thanks to Anthropic's new offering.
Companies gain the ability to operate the agent harness, along with the associated sandboxes, credentials, and session history, either within their private cloud setup or on their own physical premises. This new capability was introduced as a private beta on November 12, and interested parties must seek entry via their dedicated Anthropic account team. This development is particularly significant for autonomous agents already operating within enterprise environments, tasks such as resolving malfunctioning builds, validating new releases, managing incoming alerts, and processing accumulated bug reports. Previously, the only way customers could host these components was through the existing means. The announcement, titled "Self-hosted Claude Managed Agents keeps agent credentials in your environment," initially surfaced on Help Net Security.
Why it matters
This release allows developers to run agent workflows securely within their own infrastructure. It is significant for enterprise use cases requiring unattended agents inside corporate systems, such as bug triage or release testing.





