Claude Managed Agents

What is Claude Managed Agents?

Anthropic announced the launch of Claude Managed Agents, a new product that aims to make it easier for businesses to build and deploy AI agents. The tool offers developers out-of-the-box infrastructure to build autonomous AI systems, simplifying a complex process that was previously a barrier to automating work tasks.

In other words, it abstracts away the full engineering stack required to run agents in production.

Claude Managed Agents abstracts away the months of infrastructure work that has kept most enterprises from shipping agents in production. Users can define the agents they want to run — either by describing the agent in natural language or through a YAML file — define their guardrails, and run them on Anthropic's platform, with all of the infrastructure abstracted away.

Core Architecture

Claude Managed Agents provides the harness and infrastructure for running Claude as an autonomous agent. Instead of building your own agent loop, tool execution, and runtime, you get a fully managed environment where Claude can read files, run commands, browse the web, and execute code securely. The harness supports built-in prompt caching, compaction, and other performance optimizations for high-quality, efficient agent outputs.

The platform is built around four key concepts (from Anthropic's official docs):

The "Brain vs. Hands" Design Philosophy

Harnesses encode assumptions that go stale as models improve. Managed Agents is a hosted service in the Claude Platform that runs long-horizon agents on your behalf through a small set of interfaces meant to outlast any particular implementation. Building Managed Agents meant solving an old problem in computing: how to design a system for "programs as yet unthought of."

The architectural breakthrough was decoupling the "brain" (Claude and its harness) from the "hands" (sandboxes and tools).

Anthropic said the Managed Agents include production-grade agents with secure sandboxing, authentication, and tool execution handled for you. Agents can also run for hours in long sessions, and their progress is saved even during disconnections.

Key Capabilities

• Claude Managed Agents automates much of the work involved in state management. Tool orchestration is another task that Managed Agents promises to ease. When an agent receives a prompt, the service determines which of the tools at its disposal should be used to generate an answer. An error recovery mechanism enables agents to pick off where they left off after an outage interrupts their work.
• In research preview, the first feature enables an agent to spin up other agents when working on complex tasks.
• The other feature configures Claude to automatically refine prompt response quality. Anthropic says this capability "improved outcome task success by up to 10 points over a standard prompting loop" in internal testing.

Pricing

Pricing is straightforward.

Users pay for the models' token use, based on Anthropic's standard API pricing, with an additional $0.08 per session-hour for active runtime (measured in milliseconds).

Idle time — when an agent is waiting for your next input or a tool — does not count towards this runtime.

When the agent performs a web search, Anthropic charges an extra $10 per 1,000 searches.

When to Use It

Over 70% of companies experimenting with AI cite deployment challenges as their biggest barrier, not model performance. Anthropic is essentially offering to remove that barrier entirely with Managed Agents.

Some common use cases:

Managed Agents allows developers to create agents that can run autonomously for hours in the cloud, monitor what other Claude agents are doing, and toggle permissions that allow agents to access certain tools.

If you want to try it, log on to platform.claude.com.