Deviceways

Your assistant. Your hardware.

Let AI work with your devices

A simpler TV home screen. A router that can explain its settings. These projects let an AI assistant connect to your hardware and help you change how it works.

The connection is what makes this possible: your assistant can inspect the device, use its controls or make changes you approve. Each project below explains what you can do and how to get connected.

01

Choose a project

Find the outcome you want and check its device requirements.

02

Connect your assistant

Follow the original guide for local debugging, an SSH connection or an MCP connector.

03

Review the changes

Keep the features you use. Check the recovery steps before changing settings or packages.

A connection gives the assistant real access. The wrong change can interrupt features or leave a device needing recovery. Start with inspection and approve changes deliberately.

Projects to explore

7 projects
AI-assisted control

Let an AI assistant simplify your TV’s home screen and disable unwanted preinstalled services.

AI connects to the device

Applies toAndroid TV and Google TV with available ADB debugging. Check the contributor report for your exact model and software build.

How to get started

Open the project’s original prompt, fill in your TV model and connect from a computer on the same network. Approve debugging on the TV, then review which apps and features to keep.

Setup
AI-guided ADB changes; optional launcher installation
You’ll need
A computer, a coding assistant with local shell access and TV debugging. The launcher build has additional requirements in the guide.
Original project github.comAuthor’s AI workflow github.com

Reported devices

See all 10 reported devices →

Open a device for the reported software version and setup.

AI-assisted control

Let a local AI assistant inspect router status and change settings through MCP tools.

AI connects to the device

Applies toThe author reports specific GL.iNet models and firmware versions. See the device pages below for those environments.

How to get started

Check the project’s router and firmware requirements, install its router service, then connect your local MCP assistant through the documented SSH tunnel.

Setup
Router MCP service plus a local MCP client
You’ll need
Root SSH access, SSH keys, a supported package environment and a local MCP client. Requirements differ by firmware.
Original project · Author’s AI workflow github.com

Reported devices

Open a device for the reported software version and setup.

AI-assisted control

Let an AI assistant see your TV screen, navigate apps and inspect Android settings.

AI controls the device

Applies toAndroid devices with available ADB debugging. A TCL C6K page-display demo is documented; exact model/build coverage for the other tools is not established. Shell and uninstall tools can change the device.

How to get started

Add the local server to your MCP client, enable TV debugging and approve the connection using the original pairing guide.

Setup
Local MCP server and ADB pairing
You’ll need
Computer with Node.js 18+, Android platform-tools and a local MCP-capable assistant.
Original project github.com

Platform project. Check the original requirements for your exact model and software version.

AI-assisted control

Customize a Fire TV setup with a simpler launcher, fewer unwanted services and optional domain filtering.

AI connects to the device

Applies toPlatform: Unnamed Fire TV Edition reference; README Fire OS 7.7.1.7 / Android 9

How to get started

Use a local computer with Bash and ADB to connect to the TV and follow the project’s setup scripts. A coding assistant can use the project’s AI instructions.

Setup
Local scripts over ADB
You’ll need
A supported Fire TV setup with authorized ADB access
Original project github.comInstallation guide github.comAuthor’s AI workflow github.com

Platform project. Check the original requirements for your exact model and software version.

AI-assisted control

Ask an assistant to change volume, switch inputs or launch apps on your LG TV.

AI controls the device

Applies toNo exact model/build test matrix is supplied. Check which controls your TV supports; raw commands and picture-setting changes need care.

How to get started

Configure the local MCP server with the TV address, then approve the connection request using the TV remote. Pair each TV separately.

Setup
Local MCP server and TV pairing
You’ll need
Computer with Node.js 18+, an MCP-capable assistant and an LG webOS TV on the same network.
Original project github.com

Platform project. Check the original requirements for your exact model and software version.

AI-assisted control

Choose your TV’s home-screen app or prepare a single-app display, with an optional AI-assisted setup.

AI connects to the device

Applies toAndroid TV 7+ and Google TV families are listed, without exact model/build tests. Features differ by edition: the Google Play app excludes APK installation and debloating.

How to get started

Follow the original setup or AI-agent guide, connect over debugging and review the changes the tool proposes.

Setup
Setup app, or local MCP server over ADB
You’ll need
Android TV or Google TV with debugging; a supported host and MCP client for AI control.
Original project github.comAuthor’s AI workflow github.com

Platform project. Check the original requirements for your exact model and software version.

AI-assisted control

Choose a small home screen and turn off unwanted TV apps with an AI assistant.

AI connects to the device

Applies toPlatform: Android TV / Google TV with authorized ADB

How to get started

Enable device debugging, connect a local computer, then use the author’s README prompt and choose your apps.

Setup
Local ADB app and launcher changes
You’ll need
Authorized network ADB; local shell-capable Claude Code or Codex; physical remote. AI is optional for manual installation.
Original project github.comAuthor’s AI workflow github.com

Platform project. Check the original requirements for your exact model and software version.

Found a new way in?

Share a project that connects an assistant to a device, or a report from someone using it.

Contribute