AgentDock / Device Runtime

Give AI a secure way to work on your real devices.

AgentDock is the local runtime behind NexusDock device access. Install it on a computer, phone or server, pair the device once, and compatible AI clients can use the capabilities you expose through NexusDock.

Download AgentDock→ View on GitHub↗

What AgentDock does

NexusDock is the service that connects your AI clients, devices, memory and workflows. AgentDock runs on the device itself and exposes the local capabilities that AI can use, such as files, commands, browser automation and device-specific tools.

Runs locally

The runtime stays on the device that owns the apps, files and browser you want AI to use.

Exposes explicit capabilities

AI sees the tools and capabilities that AgentDock intentionally exposes instead of receiving unrestricted device access.

Supported platforms

AgentDock supports macOS, Windows, Linux and Android. The available capabilities depend on the platform and the permissions you choose to grant on that device.

macOS

Desktop apps, files, browser and local development workflows.

Windows

Desktop, WSL, files, commands and Windows-native capabilities.

Linux

Servers, commands, files, services and always-on automation.

Android

Mobile device actions and Android-specific runtime capabilities.

How AgentDock and NexusDock fit together

Install AgentDock

Install the official package on the device you want AI to use.

Pair the device with NexusDock

Use a one-time pairing code so the device appears in your NexusDock workspace.

Connect your AI client

Connect ChatGPT, Claude, Cursor or another compatible Remote MCP client to NexusDock.

Use the device from AI

The AI can inspect the capabilities exposed by the paired device and use the appropriate tools for the task.

Read the setup guide→ Connect an AI client→