Run Claude Managed Agents with Vercel Sandbox
You can now run Claude Managed Agents on Vercel Sandbox, enabling agent tool calls to execute on your existing Vercel infrastructure with access to your private APIs and internal services. Each agent session runs in an isolated Firecracker microVM with enterprise-grade security, credential brokering, and low-latency connectivity to your backend systems.
You can now run with .Claude Managed AgentsVercel Sandbox
Claude Managed Agents handles the model, harness, tools, and session state. Self-hosting lets you bring the execution environment, so an agent's tool calls run on your existing Vercel infrastructure with your private APIs, internal services, and customer data.
Each agent session runs in its own isolated Firecracker microVM, using the same infrastructure that powers 1B+ Vercel deployments with enterprise-grade security, availability, and performance.
What you get:
Follow the for a reference implementation and learn more in the . step-by-step guideVercel Sandbox docs
from Anthropic, including the model, harness, tools, and session stateA managed agent loop
that spawns a sandbox per sessionA Vercel function control plane
per session, with millisecond startupFirecracker microVM isolation
at the firewall, so secrets never enter the sandboxCredential brokering
with a domain allowlistDeny-by-default egress
to your private network and cloud workloadsLow-latency connectivity
Source: original entry ↗