#2828th of 54 in Assistants

orkas

Desktop app where a Commander agent delegates to specialist and CLI agents

Stars
9.2k
License
MIT
Last commit
Nov 2026
Last release
Nov 2026
Language
JavaScript
Stars, last 28 weeks+9,156
27 Apr 2026: 83 stars4 May 2026: 92 stars11 May 2026: 96 stars18 May 2026: 97 stars25 May 2026: 102 stars1 Jun 2026: 103 stars8 Jun 2026: 105 stars15 Jun 2026: 108 stars22 Jun 2026: 112 stars29 Jun 2026: 113 stars6 Jul 2026: 114 stars13 Jul 2026: 316 stars20 Jul 2026: 529 stars27 Jul 2026: 792 stars3 Aug 2026: 1,038 stars10 Aug 2026: 1,174 stars17 Aug 2026: 1,346 stars24 Aug 2026: 1,480 stars31 Aug 2026: 1,624 stars7 Sept 2026: 1,794 stars14 Sept 2026: 1,959 stars21 Sept 2026: 2,070 stars28 Sept 2026: 2,098 stars5 Oct 2026: 2,126 stars12 Oct 2026: 2,151 stars19 Oct 2026: 2,203 stars26 Oct 2026: 2,253 stars2 Nov 2026: 8,697 stars9 Nov 2026: 9,239 stars

83 on 27 Apr 20269,239 on 9 Nov 2026

From GitHub's daily star counts.

Checked on GitHub.

Overview

Orkas is a local-first Electron desktop app in which a Commander agent plans work and delegates it to 16 built-in specialists (research, video, office files, e-commerce) and to installed CLI agents: Claude Code, Codex, OpenCode, OpenClaw and Hermes-Agent. Projects hold goals, memory and files across conversations, with to-dos, auto-advance and scheduled automations. It runs from source on macOS, Windows or Linux.

Who it is for: Small teams and solo operators wanting a local multi-agent desktop workspace

Strengths

  • Orchestrates Claude Code, Codex, OpenCode, OpenClaw and Hermes-Agent under one plan
  • Projects keep memory and files; to-dos, auto-advance and scheduled automations
  • Data stays on disk; model calls go direct to your provider or local endpoint
  • 16 bundled specialist agents plus a marketplace of 30+ more

Weaknesses

  • Shared projects, cloud sync and web-app publishing only in the packaged app
  • Scheduled automations need the computer awake; missed runs catch up within 12 hours
  • Source first launch downloads Python, FFmpeg, whisper.cpp and a ~95 MB embedding model
  • Built-in Orkas model runs through a managed service and uses credits

What it needs

  • no GPU
  • Needs Node 22.12+, Python 3 and C/C++ toolchain (native-module fallback), glibc 2.34+ on Linux
  • Models: Claude, OpenAI, Gemini, DeepSeek, Kimi

What changed

From the release notes. No flag means the note does not say so, not that the upgrade is safe.

Quick answers

Is orkas open source?

Yes. Its code is on GitHub under the MIT license.

Does orkas need a GPU?

No, it runs without one.

Also in Assistants

See all 54
Also in Assistants
RankProjectScore
1ai-job-searchClaude Code workflow that searches job boards and drafts tailored CVs and cover letters71.2k starsMIT96 out of 100
2career-opsJob-search agent that scores listings, tailors CVs and drafts applications81.9k stars+8,035 (+11%) in 30 daysMIT95 out of 100
3OpenClawPersonal AI assistant reachable through Discord, Slack, Telegram, WhatsApp and more394.3k stars+4,398 (+1.1%) in 30 daysMIT93 out of 100
#4openhumanRust agent harness with desktop app, built-in memory and embeddable library42.7k stars+984 (+2.4%) in 30 daysGPL-3.093 out of 100
#5airiAI VTuber companion with VRM and Live2D avatars, voice, and game-playing Live demo ↗ (opens in a new tab)51.3k stars+1,184 (+2.4%) in 30 daysMIT91 out of 100