Description
Omnara is an open-source alternative to managed agent platforms that provides a single API for building agents that are durable, secure, and readily accessible to teams and customers. It delivers the infrastructure a production agent needs as a service: durable execution so agents run for minutes, hours, or days, survive crashes, and resume exactly where they left off, with every turn, tool call, and approval stored as durable state rather than in-process memory.
Read More
The platform provides flexible machines that agents create and destroy on demand or connect to your own devices and environments, plus controlled access through user, team, and role-based permissions, encrypted scoped secrets, approval gates for risky actions, and an append-only, queryable audit history. You define the agent's model, machines, tools, and permissions in the console or a small YAML config, and Omnara owns hosting, state, and durable execution. It is model-agnostic across OpenAI, Anthropic, OpenRouter, or any compatible endpoint, and fully self-hostable.
Omnara also serves as a command center for coding agents, letting you run and steer Claude Code and Codex from a desktop app, web, native iOS and Android apps, and Apple Watch. It supports parallel agents, worktrees, and Git integration, two-way voice conversation and dictation, and sessions that migrate to the cloud when your laptop goes offline so agent state and uncommitted changes keep running.
Security features include AES-256-GCM envelope-encrypted secrets, role-based access, platform-enforced tool policies, and immutable audit logs. The coding command center is free with unlimited sessions, while the agent platform is free to start with usage-based managed models and machines and a custom self-hosted enterprise option.
- CategoryAI Agent Builders
- Pricing ModelFreemium
- DeploymentCloud, Self-hosted
- PlatformsWeb, Windows, macOS
Omnara's Core Features
Durable agent execution that survives crashes and resumes mid-run
Flexible managed or bring-your-own machines as live tools
Role-based access, scoped encrypted secrets, and approval gates
Append-only, queryable audit history of every tool call
Model-agnostic across OpenAI, Anthropic, OpenRouter, or custom endpoints
Open-source and self-hostable in your own VPC or on-prem
Command center to run Claude Code and Codex from any device
Parallel agents, worktrees, Git, voice, and cloud session migration
How to use Omnara?
Define the agent: Use the console or a small YAML config to set the instruction, model, machines, and tools the agent may use.
Launch: Create an agent from that config and it starts working immediately with durable state.
Converse: Stream its events live, send follow-up input, and resolve the approvals and questions it raises.
For coding agents: Download the desktop app or install the CLI, sign in, and run Claude Code or Codex from desktop, web, mobile, or Apple Watch.
Omnara's Use Cases
- Production agent infrastructure
- Self-hosted agent platform
- Coding agent command center
- Remote agent control
- Governed agent access
