# Orkas

> Open-source, local-first desktop AI workforce where a Commander agent coordinates specialist agents through one chat to deliver research, documents, slides, videos and apps.

Orkas is an open-source, local-first multi-agent AI desktop client for macOS and Windows. A Commander agent takes a goal described in chat, plans it, and routes the work to specialist agents such as DeepResearcher, ContentWriter, PptMaker, ImageStudio, VideoStudio and OfficeWorker. The desktop client is released under the MIT license, and the product is aimed at indie builders, one-person businesses and small teams.

## What It Is

Orkas is a desktop workspace for directing a team of AI agents. You describe the outcome you want, attach only the relevant files or Library sources, and Commander breaks the work down, assigns it, tracks progress and verifies results. Outputs are files you can use or keep editing, such as briefs, spreadsheets, slide decks, images and videos.

## How the Workflow Works

Work is organized around tasks, Projects, Automations and reusable capabilities:

- Projects hold shared instructions, memory, agents, Library sources and to-dos for long-running goals.
- Automation runs a one-time or recurring instruction on a schedule on a chosen device.
- Agents and Skills package roles and repeatable methods, and can be created or installed from the Marketplace.
- Library processes reference documents so answers can be grounded in selected sources.
- My Apps saves eligible interactive results, such as dashboards or trackers, as reusable tools.

## Models, Agents and Connectors

You can use Orkas-managed official models or connect your own provider through OAuth or an API key, including OpenAI, Claude, Gemini, DeepSeek, Kimi and GLM, or a custom OpenAI-compatible endpoint. External CLI agents such as Codex, Claude Code and OpenCode can be called from the same conversation. Connectors and custom MCP servers give tasks access to external services, with permission modes that ask before sensitive actions.

## Local-First and Security Model

By default, configs, history, knowledge bases and generated files stay on your machine, and provider API keys are stored locally with calls going directly to the provider. Optional cloud sync and sharing, and Orkas-managed models, use Orkas services only for those selected features. Orkas says critical actions pause for confirmation and agents operate within defined access boundaries.

## Platforms and Availability

The official download offers macOS and Windows builds, while the open-source repository also supports Linux, which you build yourself. The interface offers 18 language options. The open-source desktop edition runs without an Orkas account if you add your own provider keys.

## Features
- Commander agent that plans, routes, tracks and verifies work
- Specialist agents for research, writing, slides, images, video, office documents, product development, UI design and SEO
- Projects with shared instructions, memory and to-dos
- Scheduled and recurring Automations
- Reusable custom agents and skills with Marketplace
- Library for source-grounded Q&A over reference documents
- My Apps for saving interactive results as reusable tools
- Managed official models or bring-your-own provider keys
- External CLI agent support (Codex, Claude Code, OpenCode)
- MCP servers and connectors
- Optional cloud sync and Project sharing
- Local storage, encrypted credentials and critical action confirmation
- Built-in browser for agent tasks
- 18 interface languages

## Integrations
OpenAI, Anthropic, Google Gemini, DeepSeek, Zhipu GLM, Moonshot / Kimi, MiniMax, Doubao, OpenRouter, Claude Code, Codex, OpenClaw, OpenCode, Hermes, MCP

## Platforms
WINDOWS, MACOS, LINUX, WEB, API, VSC_EXTENSION, CLI

## Pricing
Freemium — Free tier available with paid upgrades

## Links
- Website: https://orkas.ai/
- Documentation: https://orkas.ai/docs/
- Repository: https://github.com/Orkas-AI/Orkas
- EveryDev.ai: https://www.everydev.ai/tools/orkas