Anthropic has expanded its Claude Managed Agents platform with two enterprise-focused features: self-hosted sandboxes (public beta) and MCP tunnels (research preview). Self-hosted sandboxes let tool execution run on customer-controlled infrastructure via providers like Cloudflare, Daytona, Modal, and Vercel, giving enterprises control over network policies, audit logging, and data residency. MCP tunnels allow Managed Agents to connect to private MCP servers without exposing them to the public internet — a lightweight gateway establishes an outbound encrypted connection to Anthropic infrastructure instead of requiring inbound firewall rules. Together, these features address a key enterprise blocker: running autonomous agents inside a security perimeter without lengthy security reviews.
