skalesapp/skales

Personal AI desktop agent for Windows, macOS, Linux, Android & iOS. Set a goal, it works on its own. Teams (pair two desktops, agents + humans), Agent2Agent, Workflows, Codework, multi-agent orgs, desktop + browser automation. 15+ AI providers, BYOK. No Docker, no terminal. Agent Skills (SKILL.md). Migration importer. Recurring autonomous tasks.

Skales – a private AI desktop / mobile agent

What it is – Skales is a locally‑run AI assistant that can chat, write code, design graphics/video, manage calendars, and automate multi‑step tasks. It runs as a native application on Windows, macOS, Linux, Android and iOS, and can be paired with a phone so you can drive the desktop from anywhere. The core product is closed‑source (BSL 1.1) but the GitHub repo hosts the historic v7 source snapshot, release binaries, changelog and issue tracker.

Key capabilities

  • Goal‑oriented automation – /goal turns a natural‑language request (e.g., "build me a trading bot") into a background workflow that continues even when the app is closed. The agent plans, executes, and only asks for confirmation on consequential actions such as sending email.
  • Code‑centric workspace – /code binds a folder to a chat, shows inline git diffs, lets you approve each edit or run in full‑auto mode, and can commit/push using your own Git identity.
  • Studio (design & media) – A “Flow” UI where you describe a visual or audio brief and the agent generates slide decks, wireframes, 3‑D three.js scenes, images, videos, motion graphics, or music. Generated assets are saved as real files; you can edit, re‑prompt, and export MP4/PNG/etc.
  • Multi‑modal chat – Supports text, voice (local TTS, OpenAI, ElevenLabs, Azure), image/video rendering, Mermaid diagrams, LaTeX math, and rich card‑based search results. Prices for LLM calls are shown live, with per‑task ceilings.
  • Cross‑device access – Pair a phone via QR code; the mobile app can issue the same commands, view the desktop UI, or run as a standalone agent.
  • Memory & continuity – Short‑ and long‑term memory across chats, phone, WhatsApp/Telegram integrations, and optional Obsidian vault linking. Memory can be turned off for full privacy.
  • Pluggable LLM providers – Works out‑of‑the‑box with a free “Skales IQ” model (no API key), and can connect to >15 external providers (OpenAI, Anthropic, Groq, etc.) or run fully offline with Ollama.
  • Privacy‑first – All data stays on the user’s machine unless you explicitly enable cloud services; wake‑word voice detection runs locally.

Installation – One‑click installers are provided for each platform (EXE, DMG, AppImage). No Docker, no terminal, no Python environment required.

License – The binary product is free for personal use under the Business Source License 1.1. The repository’s source tree is a frozen historical snapshot and is not the code that ships.

Typical users – Anyone who wants an AI assistant that can actually manipulate files, run scripts, and create media without sending their data to the cloud – from hobbyists building small games to professionals automating repetitive office work.

Links

Related

  • Project
    Orkas-AI/Orkas

    A local-first multi-agent desktop app where a central Commander agent coordinates a team of specialist AI agents to complete complex goals from a single chat.

  • Project
    hAcKlyc/MyAgents

    An open-source desktop workspace for personal AI agents that integrates local files, terminals, and a task scheduling system to move beyond simple chat interfaces.

  • Project
    openteams-lab/openteams

    A local-first AI desktop app that provides a shared workspace for multiple coding agents to collaborate, plan, and execute tasks without interfering with each other.

  • Project
    pmbstyle/Alice

    An open-source AI desktop companion that integrates voice interaction, system-level computer control, and long-term memory to assist with daily tasks and creative work.

  • Project
    holaboss-ai/holaOS

    An agentic workspace that integrates AI agents side-by-side with real applications and chat tools, featuring local-first memory and shared tool access.