Orkas
Orkas is an open-source, local-first multi-agent desktop app that lets you command a team of AI agents from a single chat interface. It solves the problem of coordinating multiple specialized AI tasks by using a Commander agent to plan and delegate work to nine built-in specialist agents. It is for developers and professionals who want to automate complex workflows involving research, content creation, coding, and file generation while keeping data and API keys local.
✨ Key features
- Commander agent plans and coordinates specialist agents.
- Nine specialist agents ship with the app.
- Local-first storage: data and keys stay on your disk.
- No vendor lock-in: mix AI providers across agents.
- Agents have private skills and memory, improve over time.
- Can drive external CLI coding agents like Claude Code.
🎯 Use cases
- Automate recurring reports and market research.
- Turn a product spec into dev tasks.
- Chat with documents and run local data analysis.
- Produce code, research, video, and slide decks from one chat.
📦 Installation
🧰 Requirements: Requires Node 20+ for source runs; packaged installers for macOS and Windows 10+; glibc 2.34+ Linux x64/arm64 for source. Bring your own model API keys or use local model endpoints.
git clone https://github.com/Orkas-AI/Orkas.git
cd Orkas
./run.sh # macOS / Linux
run.cmd # Windows
🚀 Usage
# After installation, launch the app and open Settings → AI Providers to add an API key.
# Then describe a goal in the chat, e.g., "Research the top 5 competitors, write it up, and turn it into a deck."
# The Commander will plan and delegate to agents like DeepResearcher, ContentWriter, and PptMaker.
⚠️ Good to know
Linux source runs require glibc 2.34+; Alpine and musl-based distributions are not supported. The packaged desktop app asks for sign-in on first launch, but the source build has no account layer.
❓ FAQ
Is Orkas a local LLM?
No. Orkas runs on your machine but calls models through your own API keys or a local model endpoint. It orchestrates agents and tools, not a model itself.
Can I use local models like Ollama or LM Studio?
Yes. Add them under Settings → AI Providers → Custom (OpenAI-compatible) and point the base URL at your local endpoint. You can mix local and hosted models across agents.
Where are my API keys and data stored?
On your disk. Conversations, files, knowledge bases, agents, and keys stay local; model calls go directly from your machine to the provider and are never proxied or archived by Orkas.
Does Orkas work offline?
The app is fully offline-capable; only model calls need network. Point agents at a local model endpoint to run without the cloud.
📊 Repository
🤖 Overview, features, install steps and FAQ were generated from the project's README on Sep 4, 2026. Always check the original source before running commands.