7,363 open-source and SaaS tools, with GitHub stats refreshed every day.

Open Managed Agents

Open source

Open-source, self-hostable runtime for AI agents that implements Claude Managed Agents and OpenAI Agents API-compatible endpoints with durable sessions.

Open-source alternative to

openma.ai
Open Managed Agents homepage screenshot
GitHub stars
313
Last commit
yesterday
Repository age
6 months
Version
@openma/sdk@1.0.0
Licence
Apache-2.0
Self-hosted
Yes

About Open Managed Agents

Open Managed Agents (OpenMA) is an open-source platform for running AI agents under your own control. It reimplements the Claude Managed Agents API and also exposes the OpenAI Agents API, so existing clients and the official OpenAI SDK can point at it. Agents run with durable sessions, sandboxed tools, memory, encrypted credentials and crash recovery, and you bring your own model keys.

There are two ways to run it. The Node host uses SQLite or Postgres with local file storage and a choice of sandboxes such as LiteBox, Daytona, E2B or BoxRun, and starts with docker compose. The Cloudflare host runs on Workers, Durable Objects and Containers with D1, KV and R2, and deploys with wrangler. Both expose the Claude-compatible agents and sessions API, a console UI and the OpenAI-compatible endpoint, and support MCP, private tools and encrypted vaults.

OpenMA is written in TypeScript and licensed under Apache-2.0. A hosted version is available at app.openma.dev for teams that prefer not to operate it, while the self-hosted options suit on-premises or data-resident deployments. It targets teams that want an alternative to Claude's managed agent service while choosing their own models and infrastructure.

Key features

  • Claude Managed Agents compatible API
  • OpenAI Agents API compatible endpoint
  • Durable sessions with crash recovery
  • Sandboxed tool execution
  • Encrypted credential vaults
  • MCP and private tool support
  • Node.js or Cloudflare deployment
  • Bring your own model keys

Good fit for

  • →Running agents on-premises with your own model keys
  • →Replacing a managed agent service with a self-hosted one
Built with
TypeScript
Tags
ai-agents
agent-platform
self-hosted
cloudflare
mcp
byok
typescript
sandbox

Open Managed Agents: questions and answers

What is Open Managed Agents used for?
Open Managed Agents is an open-source, self-hostable runtime for AI agents that implements Claude Managed Agents and OpenAI Agents API-compatible endpoints with durable sessions. It is a good fit for running agents on-premises with your own model keys and replacing a managed agent service with a self-hosted one.
Is Open Managed Agents open source?
Yes. Open Managed Agents is open source under the Apache-2.0 licence. Its source code is on GitHub at openma-ai/open-managed-agents and is written mainly in TypeScript.
Is Open Managed Agents free?
Yes. Open Managed Agents is open source, so the software itself is free to use. A managed cloud version is also available.
Can I self-host Open Managed Agents?
Yes. Open Managed Agents can be self-hosted on your own server or infrastructure.
What is Open Managed Agents an alternative to?
Open Managed Agents is an open-source alternative to Claude. Other open-source alternatives to Claude include Ollama, Open WebUI and oh-my-claudecode.
Is Open Managed Agents actively maintained?
Yes. The most recent commit to Open Managed Agents was on 2 October 2026. The project has 313 stars on GitHub.

Open-source alternatives to Open Managed Agents

See all

SaaS alternatives to Open Managed Agents

See all