Claude Managed Agents
Build and deploy long-running, secure agents tuned to Claude’s latest models, without managing infrastructure.
Secure infrastructure
Credentials are kept out of the sandbox, encryption is built in, permissions can be scoped for each agent, and state saves automatically.
Agent observability
Session tracing records what every agent did and why, with built-in analytics to improve agent performance.
Optimized for Claude
The harness is tuned to Claude, so tool calls, context handling, and retries match how the model behaves.
Memory
Every session makes the next one better
Claude writes what it learns to memory, and later sessions read from it.
Multiagent orchestration
A lead agent assigns work to specialists
Each specialist has its own model, prompts, and tools. They share a filesystem to collaborate in.
Outcomes
Claude grades its own work, then fixes it
The grader scores each output and sends it back for revision until it meets your bar.
Vaults
Credentials stay out of the sandbox
The vault adds your secrets at the network boundary, so the agent only handles a placeholder.
Scheduled deployments
Agents run on a schedule you set
Runs at the times you set. Memory persists between runs.
Self-hosted sandboxes
Agents’ tools can run in your environment
Tools run and execute on files in your infrastructure, using clients preconfigured to your specifications.
Code migration
Modernize legacy systems with agents that rewrite codebases in a new language, refactor logic, and replace outdated APIs.
Security remediation
Deploy agents to autonomously manage incident response, patch vulnerabilities, and protect against novel threats by writing detection rules.
Experimental research
Speed up research and design with agents that run experiments, analyze the results, and iterate progressively until the target is reached.
Compliance audits
Run compliance audits with agents that gather evidence, test controls, and report what needs fixing.
Describe your agent, or start from a template with the tools, connections, and permissions already set.
Deep researcher
Conducts multi-step web research with source synthesis and citations.
Field monitor
Scans software blogs for a topic and writes a weekly what-changed brief.
Contract tracker
Extracts clauses, sets deadline reminders, and tracks obligations from a file ID or link.
Support agent
Answers customer questions from your docs and knowledge base, and escalates when needed.
Sprint retro facilitator
Pulls a closed sprint, synthesizes themes, and writes the retro doc before the meeting.
Data analyst
Loads, explores, and visualizes data; builds reports and answers questions from datasets.
Incident commander
Triages an alert, opens an incident ticket, and runs the incident room.
Support-to-eng escalator
Reads a support conversation, reproduces the bug, and files a linked issue with repro steps.
Structured extractor
Parses unstructured text into a typed JSON schema.
“Atlassian helps enterprises orchestrate work across humans and agents. With Claude Managed Agents, we can build agents for developers directly into the workflows teams already rely on in weeks instead of months, so customers can assign tasks right from Jira. Managed Agents handles the hard parts like sandboxing, sessions, and scoped permissions, which means our engineers can spend less time on infrastructure and more time building great features for our end users.”
“Claude Managed Agents made it 3x faster to build a production-ready meeting prep agent. We went from idea to shipping in a matter of days. Our agent researches every participant ahead of a meeting to surface what matters for moving the conversation forward. Custom tools let us feed in our own calendar and contacts data, MCP made it simple to connect external systems like meeting notetakers, CRMs, etc., and the managed harness handled the heavy lifting, including sandboxed execution and built-in web search. Letting us focus on building the product, not the infrastructure.”
“Turns out telling developers what’s wrong with their code isn’t enough: they want you to fix it too. Customers can now go from Seer’s root cause analysis straight to a Claude-powered agent that writes the fix and opens a PR. We chose Claude Managed Agents because it gives us a secure, fully managed agent runtime, allowing us to focus our efforts on building a seamless developer experience around the handoff. Managed Agents not only allowed us to build the initial integration in weeks instead of months, but has also eliminated the ongoing operational overhead of maintaining bespoke agent infrastructure.”
“Claude Managed Agents dramatically accelerated our development of Asana AI Teammates — helping us ship advanced capabilities faster — and freeing us to focus on creating an enterprise-grade multiplayer user experience.”
“We want Notion to be the best place for teams to work with agents and get things done. We integrated Claude Managed Agents, which can handle long-running sessions, manage memory, and deliver high-quality outputs over time, to make that possible. Our users can now delegate open-ended, complex tasks, everything from coding to generating slides and spreadsheets, without ever leaving Notion.”
“With Claude Managed Agents, our power users become like Galileo, contributing across domains far beyond a single specialty or discipline. We deploy each specialist agent within a week, managing long-running tasks across engineering, product, sales, marketing, and finance, generating apps, proposal decks, and spreadsheets in sandboxed environments. As agents become more capable, Managed Agents lets us scale safely without building agentic infrastructure ourselves, so we can focus entirely on democratizing innovation across the company.”
“We built a cross-account agentic search system on Claude Managed Agents that lets sales teams ask things like which accounts to reach out to today. Since customers want those answers refreshed regularly, we replaced the scheduling infrastructure we'd built ourselves with scheduled deployments, which greatly simplified our stack and improved our product cycles.”
“Spiral’s writing agent runs on multiagent orchestration and outcomes. A lead agent on Haiku fields requests and hands the drafting to Opus subagents, which run in parallel when a user wants multiple drafts. Writing quality is the core of what we do, so every draft is scored against a rubric of Every’s editorial principles and the user’s voice. We only return the ones that clear the bar.”
Claude Managed Agents is a suite of composable APIs for building and deploying cloud-hosted agents at scale. It pairs an Anthropic-managed harness with production infrastructure for state, memory, permissions, and scheduled execution.
The Agent SDK runs locally on your own machines. Claude Managed Agents gives you the hosted infrastructure to run agents in the cloud, with sessions, sandboxes, and permissions managed for you.
Each agent’s code runs in its own isolated sandbox on the Claude Platform.
If you need tools, files, and credentials to stay in your environment, self-hosted sandboxes keep them in your cloud.
Credentials are stored in a vault and added at the network boundary, so the agent and its code only ever see a placeholder. Instructions are kept outside the sandbox too, so nothing the agent runs can reach them.
You can delete sessions through the API and use your own encryption keys for stored data.
For each tool, you choose whether it runs automatically, always waits for approval, or is checked first and then run, denied, or paused for a person to review. Tools from connected MCP servers ask for approval by default.
You can also limit networking with domain allowlists, so the sandbox, web search, and web fetch only reach the hosts you name.
Claude Managed Agents is currently available on Claude Platform on AWS.
No, Claude Managed Agents is not currently HIPAA eligible.
Anthropic charges token rates plus $0.08 for each hour a session runs. There are no platform or seat fees, and it counts towards commits.