claudekit / updates / self-hosted-claude-managed-agents
[ NEW · ]

Self-hosted Claude Managed Agents — operate Managed Agents inside your security perimeter (private beta)

Self-hosted Claude Managed Agents is now available in private beta. You host the Managed Agents service and its sandboxes on your own infrastructure, in a cloud account or an on-premise data center, store agents, sessions, and session history in your own database, and choose where requests to Claude are sent: the Claude API, Amazon Bedrock, Claude Models on Google Cloud (Google managed), Microsoft Foundry, or your own model gateway.

Official announcement →

This article is a summary based on official documentation.

Overview

Anthropic introduced self-hosted Claude Managed Agents on November 12, 2026, available in private beta. Until now, Anthropic ran the Managed Agents service, and self-hosted sandboxes let you move only the sandboxes where an agent’s tools run. Self-hosted Managed Agents extends that: every part of the service runs in your own cloud or on-premise, including the harness, sandboxes, credentials, and session history, and you choose where requests to Claude are sent.

Key features

  • You host the service and the sandboxes

    With self-hosted sandboxes, the service, including the storage for agents, sessions, and session history, stayed on Anthropic’s infrastructure. Self-hosted Managed Agents moves it to you. One or more servers, their database, and your sandboxes make one deployment:

    ComponentWhat it does
    Server (cma-server)Serves the API, runs the agent loop, and stores agents, sessions, and each session’s history. It calls your inference provider whenever the agent needs the model, and never runs an agent’s shell or file tools
    DatabasePostgreSQL 16 or later. Without a database, the server keeps its data in memory, and a restart erases it
    WorkersTake work from the server and run the agent’s shell and file tools in your sandboxes. A worker is an ant beta:worker command or an SDK worker
    Console on your serverA webpage at the server’s address, where people sign in through your identity provider to create agents and run sessions
    Admin CLI (cma-server-admin)Creates environments and their environment keys. Runs on Linux (x86-64 and Arm) and macOS on Apple silicon
  • Your choice of inference provider

    With cloud sandboxes and self-hosted sandboxes, the model runs at Anthropic. Your server can send them to the Claude API (the default), Amazon Bedrock, Claude Models on Google Cloud (Google managed), Microsoft Foundry, or your own model gateway. If your organization has zero data retention (ZDR) with Anthropic, it applies to Claude requests as it does for any Messages API call.

  • The same API as the hosted service

    Your server has the same API as the hosted service, so the SDK and the ant CLI work with it, and the rest of the Managed Agents documentation applies. The “Differences from the hosted service” page lists the exceptions.

  • Features that only your server has

    Beyond a choice of inference provider, a self-hosted server adds single sign-on through your OpenID Connect identity provider, your own secret manager for vault credential secrets, and agent identity, where the server signs a short-lived token and sends it with an agent’s calls to an MCP server.

Notes

  • Private beta, access by request — Contact your Anthropic account team to request access to the download, then follow “Get started with self-hosted Claude Managed Agents” to start a server and run a first session. The announcement doesn’t list pricing.
  • You own operations and security — You run and secure a server, its database, and your sandboxes, and you install each release yourself. The agent’s bash tool is a real shell with the worker process’s permissions and network access, and the worker doesn’t isolate it. If you start a worker yourself, run it in a disposable container or VM that holds nothing you need to protect.
  • Some hosted features aren’t available — Cloud environments (only self_hosted environments exist), the Messages API and Models API, file session resources and the Files API, mcp_oauth vault credentials, the prebuilt docx, pdf, pptx, and xlsx skills, scheduled deployments, dreams, and webhooks.
  • Nothing is copied from the hosted service — Session history included. Create your agents, skills, vault credentials, and memory stores again on your server.
  • Some features need a server setting — MCP servers and Git hosts need an allowlist entry, and the web_search and web_fetch tools are off by default.
  • What leaves your network — For each model call, the server sends the conversation so far, including tool inputs and tool results, to your inference provider. The “Data and access” page lists every connection the server makes.
  • If you already run self-hosted sandboxes — The worker software is the same, so your workers can connect to your server. A worker that starts on a webhook doesn’t work here.
§ 4

Frequently Asked Questions

frequently asked
§ 4.1
What is self-hosted Claude Managed Agents?
It is Managed Agents on your own infrastructure, which can be a cloud account or an on-premise data center. You host the service (the API, the agent loop, and the storage for agents, sessions, and session history) and the sandboxes. The model runs at Anthropic or at a provider that you choose.
§ 4.2
Is it available now?
Self-hosted Claude Managed Agents was announced in private beta on November 12, 2026. To request access, contact your Anthropic account team.
§ 4.3
How is it different from self-hosted sandboxes?
With self-hosted sandboxes, you host only the sandboxes where an agent's tools run, and Anthropic runs the service. Self-hosted Managed Agents extends that: you host the service too, and you can choose the inference provider. If you only need tools to run on your infrastructure, the docs recommend self-hosted sandboxes.
§ 4.4
Where can model calls go?
By default, the server sends model calls to the Claude API. It can instead send them to Amazon Bedrock, Claude Models on Google Cloud (Google managed), Microsoft Foundry, or your own model gateway.
§ 4.5
Does everything from the hosted service work on a self-hosted server?
The API is the same, with exceptions. Cloud environments, the Files API, the prebuilt docx, pdf, pptx, and xlsx skills, scheduled deployments, dreams, and webhooks aren't available. Nothing is copied from the hosted service, session history included, so you create your agents, skills, vault credentials, and memory stores again on your server.
§ 4.6
Where can I find the official announcement?
See the announcement at claude.com/resources/articles/self-hosted-claude-managed-agents and the Self-hosted Claude Managed Agents docs at platform.claude.com/docs/en/managed-agents/self-hosted.