Anthropic adds self-hosted sandboxes and MCP tunnels to Claude Managed Agents
Anthropic just gave enterprises what they've been asking for: Claude agents that run on their own servers. Tool execution moves into your infrastructure. The agent logic stays with Anthropic.
Why it matters
Anthropic is addressing enterprise security and data sovereignty concerns by allowing companies to self-host tool execution for Claude Managed Agents while maintaining control of the agent orchestration layer. This is a critical deployment unlock for regulated industries and large orgs requiring on-premises compute.
The key facts
5 to knowSelf-hosted sandboxes now available for Claude Managed Agents
MCP (Model Context Protocol) tunnels enable enterprise infrastructure integration
Tool execution can move to customer infrastructure
Agent control remains with Anthropic platform
Addresses data sovereignty and security requirements for enterprise deployment
Go to the source
The Decoderthe-decoder.com
Publisher excerpt: Anthropic is expanding Claude Managed Agents with self-hosted sandboxes and MCP tunnels. Companies can now move their AI agents' tool execution into their own infrastructure. But Anthropic isn't handing over full control of the agent itself.