Anthropic and Cloudflare Partner to Bring Claude Managed Agents to Cloudflare Infrastructure
Key Takeaways
- ▸Claude Managed Agents can now run on Cloudflare infrastructure, offering developers infrastructure flexibility previously unavailable
- ▸The integration includes security enhancements (customizable proxies, credential injection), observability (sandbox metrics and logs), and private service connectivity
- ▸Cloudflare's recent investments in Sandboxes, Agents SDK, Browser Run, and Dynamic Workers create a comprehensive platform for running agents at scale
Summary
Anthropic and Cloudflare have announced a new integration that allows developers to run Claude Managed Agents on Cloudflare's infrastructure. The partnership follows Cloudflare's recent expansion of its Developer Platform with sandbox environments, an Agents SDK, and browser automation tools. Through this integration, developers can leverage Anthropic's agent orchestration on Cloudflare's secure, scalable infrastructure—a concept Anthropic describes as "decoupling the brain from the hands," where Anthropic's platform handles the agent loop while Cloudflare manages code execution and infrastructure.
The integration includes a default deployment template that provides enhanced security through customizable proxies, comprehensive sandbox control with metrics and logging, lightweight sandboxes that boot in milliseconds, and private service connectivity for internal resources. Additional features include browser control with session recording, email capabilities for agents, and the ability to extend agents with custom tools. The setup is designed to take just minutes, with developers able to fork and customize the provided repository as needed.
This partnership addresses a key limitation of Claude Managed Agents—previously, the entire stack ran on Anthropic-provided infrastructure. The new integration gives developers the flexibility to choose Cloudflare for better control over security, compliance, and performance requirements. As agents become more prevalent in enterprise workflows, the ability to decouple agent logic from execution infrastructure represents a significant step forward in making agent deployment more flexible and secure.
- The deployment model decouples the agent 'brain' (Anthropic's orchestration) from the 'hands' (Cloudflare's execution infrastructure)
Editorial Opinion
This partnership signals a maturation in enterprise agent deployment. By allowing Anthropic's orchestration engine to run on third-party infrastructure, both companies acknowledge that no single platform will dominate—developers need choice and flexibility. Cloudflare's aggressive expansion into agent infrastructure, combined with this integration, positions the company as a serious contender for agent workloads, while Anthropic gains distribution through a trusted infrastructure partner.
