Anthropic opens self-hosted Claude Managed Agents in private beta

Anthropic says Managed Agents, its suite of APIs for building and deploying autonomous agents, can now run entirely in a customer's own cloud or on-premise environment. Until now Anthropic ran the service, with self-hosted sandboxes as an option for teams that wanted their code to stay in their own environment.

With self-hosting, the customer operates the Managed Agents server, its database and the workers where agents run shell and file tools. That covers the harness, sandboxes, credentials and session history. Developers build agents with the same core APIs as the hosted service.

• Model requests: they go where the customer chooses: Anthropic directly, Amazon Bedrock, Google's Agent Platform, Microsoft Foundry or the customer's own gateway. Anthropic says an organization's zero data retention applies to the requests sent to Claude as with any Messages API call.
• Credentials: Anthropic says vaults keep them out of the sandbox, with a proxy outside it adding the credential to MCP and supported API calls. Customers can also keep credentials in their own secrets manager.
• Network and audit: each environment can be limited to an allowed list of hosts, and every step and tool call is recorded and exportable over OpenTelemetry.

Access is by request through an Anthropic account team.