Backed byYCombinator

The open-source alternative to Claude Managed Agents

A single API for deploying agents that are durable, secure, and accessible to your customers and team.

Start buildingTalk to a founder
npx omnara agents launch \
  --name "General agent" \
  --source '{  }' \
  --message "What tools do you have access to?"
Or prompt your agent:
Read omnara.com/SKILL.md to set up my agent

Deploy Agents with Omnara

Companies building agents have to choose whether to use closed platforms like Claude Managed Agents, or to spend months of engineering time to build their specialized internal agent platforms, like Stripe, Sentry, and Harvey. With Omnara, you can spend time building your product instead of fighting closed infrastructure.

Omnara exposes a stateful, open-source agent API so you can build:

Agents in your product's dashboard1/4

Features

Whether you're building internal agents for your team or agentic experiences for your customers, Omnara provides everything you need to build agents tailored to your use case.

Durable Agents

Omnara's Agents preserve their state so they can run forever and recover from even the worst failure scenarios.

Stream the agent timeline

Computer-use

Give agents managed sandboxes they create and destroy on demand, or connect your own devices and environments.

Connect machines and pools

Access Controls

Organize work into teams and projects with role-based access for people and API keys. Store credentials as encrypted, scoped secrets, require human approval for risky tool calls, and audit every action in an append-only queryable history.

Set tool permissions

Agent-ready APIs

Launch, steer, and observe agents from your own code. The REST API, TypeScript SDK, and CLI all drive the same control plane, so your application owns the whole loop.

Explore the API

BYO model

Bring your own API keys and use any model behind a compatible endpoint — OpenAI, Anthropic, OpenRouter, LiteLLM, or models you host yourself.

Add a model provider

Integrations

Connect MCP servers and load reusable skills to integrate with external data. Interact with your agents through the first party Slack integration.

Browse integrations

Frequently asked questions.

Everything you need to know about Omnara, self-hosting, and what you can build.

What is Omnara?

Omnara is an open-source API for running user-facing agents. The Omnara API handles execution and state while you choose the models, tools, machines, and how users interact with each agent.

Do I need to run any infrastructure?

No. Omnara provides a hosted cloud where the infrastructure (sandboxes, model providers, hosting) is managed for you. You can also self-host Omnara.

Which models are supported?

Omnara is model-agnostic and supports OpenAI, Anthropic, OpenRouter, or any compatible endpoint, including models you host yourself.

Is it really open source? Can I self-host?

Yes. Omnara is available under the Apache 2.0 license at github.com/omnara-ai/omnara. Omnara Cloud adds managed services. Self-host in your own VPC for full control, or start on the managed cloud and move later.

Do I own my data?

Yes, on the managed cloud, all data is queryable via the API. For scenarios requiring increased data security, you can always use or migrate to the self-hosted version of Omnara.

Does Omnara support permissions?

Tool policies are enforced by the platform. For risky actions, you can configure Omnara to pause for approval.

We handle the infrastructure.

You build the agent.

Start buildingTalk to a founder