openyak/openyak

Open-source local-first AI agent for desktop work. No account, no telemetry: use local models with Ollama/Rapid-MLX or bring your own provider key.

What it solves

OpenYak provides a local-first agent runtime that allows users to run AI agents directly on their own computers. It eliminates the need for hosted workspace accounts and cloud-only dependencies, allowing agents to interact with local files, native desktop applications, and a managed browser while keeping data, memory, and workflow state on the device.

How it works

OpenYak operates as a desktop application (built with Tauri, Next.js, and FastAPI) that connects to either local model providers (like Ollama or Rapid-MLX) or cloud providers via API keys (BYOK). It features two specialized workspaces:

  • Native Computer: Interacts with macOS and Windows apps using accessibility elements and coordinate fallbacks.
  • Managed Browser: A dedicated browser profile for navigating websites using page elements and screenshots.

It employs a durable agent runtime that manages persisted state, tool execution with argument validation, and a layered permission system to ensure safety and user control. Users can "take over" control of the workspace from the agent in real-time and then return it back to the agent.

Who it’s for

It is designed for professionals who need an AI agent that can handle complex office workflows—such as synthesizing multiple documents (DOCX, XLSX, PPTX, PDF) into briefs or action plans—and those who prioritize privacy and local control over their AI tools.

Highlights

  • Local-First Architecture: No account required; files and conversations stay on thedevice.
  • Shared Control: Live view of agent actions with the ability for users to manually intervene and return control.
  • Flexible Model Boundary: Supports local models (Ollama, Rapid-MLX) and a wide array of cloud providers.
  • Ultra Agent Swarm: Ability to launch parallel child-agents for complex tasks, synthesizing results into a parent session.
  • Artifact Workspace: A dedicated panel for creating reusable Markdown briefs, tables, and diagrams.
  • Remote Access: Secure mobile access to the desktop agent via QR code and Cloudflare Tunnel.

Related

  • Project
    hAcKlyc/MyAgents

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

  • Project
    AIDotNet/OpenCowork

    OpenCowork is an open‑source desktop (Electron + .NET) platform that lets LLM‑based agents work directly on your local files, shell, and repositories. It supports multi‑agent teams, integrates with many chat services, offers five interaction modes (chat, clarify, cowork, code, acp), and provides a rich toolbox (file/shell access, browser automation, task/cron management, CodeGraph indexing, skill extensions). The same native .NET worker powers both the GUI and a terminal‑based CLI, keeping a single authoritative agent loop. Quick start requires Node ≥ 18 and .NET 11 SDK; builds are available for Windows, macOS, and Linux. Use cases include autonomous coding, scheduled ops, research/report generation, remote SSH tasks, and media‑pipeline design. Licensed under Apache 2.0.

  • Project
    ThinkInAIXYZ/deepchat

    DeepChat is an open‑source, cross‑platform desktop client that lets you run LLM‑based agents locally. It unifies cloud and Ollama models, records every interaction with a recoverable “Tape” system, supports reusable Skills, ACP‑compatible agents, and remote control via Telegram, Discord, Feishu, QQBot and WeChat. The app offers rich markdown/code rendering, built‑in web‑search extensions, strong MCP tooling, and privacy‑focused storage, all under an Apache‑2.0 license.

  • Project
    lencx/Minke

    A native desktop workspace for DeepSeek Harness that integrates chat, files, terminal, and web tools into a local-first environment for agentic work.

  • Project
    open-jarvis/OpenJarvis

    A framework for local-first personal AI agents that allows users to run intelligence on their own devices, reducing reliance on cloud APIs.