Orkas
Orkas is an open-source, local-first multi-agent desktop client for macOS and Windows. A Commander coordinates specialist agents, projects, scheduled automations and connectors to turn goals into finished work.
Visit Orkas-AI/OrkasOverview
Orkas is an open-source, local-first multi-agent AI desktop client. A Commander turns goals into to-dos and coordinates specialist agents, with projects, scheduled automations and connectors, to complete complex work together.
Key Features
- Commander coordinates 16 built-in specialist agents plus supported coding agents to split and combine complex work.
- Projects keep goals, rules, memory and files so new conversations continue prior context.
- To-dos, Auto-advance and scheduled automations help work keep moving between check-ins.
- Local-first storage keeps files, chats and memory on the user's computer, with optional cloud sync.
- Model choice includes managed official models or user-provided providers through OAuth or an API key.
- MCP tool connections and shared projects support collaboration with external tools and teammates.
Use Cases
- Research markets and turn findings into reports, slides or launch videos.
- Create marketing content, images, videos and editable presentations.
- Develop and fix software with UI design, feature code and tests.
- Run recurring business operations such as weekly briefs, client updates and store reviews.
Getting Started
- Download Orkas for macOS or Windows; the app is free and includes 500 credits for managed official models.
- Official models work out of the box, or connect your own model provider and local agents.
- Use projects to set a goal, then let the Commander coordinate agents and keep work moving.
Deployment & Requirements
- Official downloads are available for macOS and Windows.
- The open-source repository also supports Linux, which users build themselves.
- Orkas is a desktop client rather than a server; self-hosting means running the app on your own machine.
Before You Adopt
- License: MIT. Review its terms before using, modifying, or distributing the project.
- By default, local data and user-provided keys stay on the user's machine, while selected managed or cloud features use Orkas services.
- Managed models and optional cloud features depend on Orkas services when those features are selected.
- Self-hosting Orkas does not mean deploying a team server; it means running the desktop app on your own machine.
- Published web apps have limitations and cannot directly use desktop AI, local files or connectors.
- Credits apply only to the Orkas-managed model service; bring-your-own-provider usage is billed by the provider.