---

name: google_tv description: Control the Living Room Chromecast with Google TV via ADB author: Anthony

Google TV Control

Use this skill when I ask to control the TV, play shows, or check if the TV is online.

Setup

This skill uses a local Python virtual environment to communicate with the TV via ADB.

• Always use the skill-specific venv at ./.venv/bin/python3 (not the workspace venv folder). Activate or call that python explicitly when running the script.
• The skill requires a few Python packages at runtime; ensure the venv has at minimum: requests and beautifulsoup4 (bs4). To install into the skill venv run:

./.venv/bin/python3 -m pip install -r requirements.txt

(See requirements.txt in the skill folder.)

Capabilities

This skill provides a small CLI wrapper around ADB to control a Google TV device. It exposes the following subcommands:

• status: show adb devices output
• play <query_or_id_or_url>: play content. Prefer providing a YouTube video ID or a provider URL/ID.
• pause: send media pause
• resume: send media play

Usage examples

./.venv/bin/python3 google_tv_skill.py status --device 192.168.4.64 --port 5555

./.venv/bin/python3 google_tv_skill.py play "7m714Ls29ZA" --device 192.168.4.64 --port 5555

./.venv/bin/python3 google_tv_skill.py pause --device 192.168.4.64 --port 5555

Device selection and env overrides

• You can pass --device (IP) and --port on the CLI.
• Alternatively, set CHROMECAST_HOST and CHROMECAST_PORT environment variables to override defaults.
• If you provide only --device or only --port, the script will use the cached counterpart when available; otherwise it will error.
• The script caches the last successful IP:PORT to .last_device.json in the skill folder and will use that cache if no explicit device is provided.
• IMPORTANT: This skill does NOT perform any port probing or scanning. It will only attempt an adb connect to the explicit port provided or the cached port.

YouTube handling — direct intent only (preferred)

• If you provide a YouTube video ID or URL, the skill will launch the YouTube TV app directly via an ADB intent restricted to the YouTube TV package (no Assistant/UI fallback).
• The skill attempts to resolve titles/queries to a YouTube video ID using the yt-api CLI (on PATH) and falls back to /Users/anthony/go/bin/yt-api if needed. If ID resolution fails, the skill will report failure rather than attempting UI search.
• You can override the package name with YOUTUBE_TV_PACKAGE (default com.google.android.youtube.tv).

Tubi handling

• If you provide a numeric Tubi ID the skill will attempt the tubitv:// intent and then a VIEW https intent for a Tubi URL, both restricted to the Tubi package.
• If you provide a Tubi https URL, the skill will send a VIEW intent with that URL (restricted to the Tubi package).
• If the canonical Tubi https URL is needed, the assistant can look it up via web_search and supply it to this skill.
• You can override the package name with TUBI_PACKAGE (default com.tubitv).

Pause / Resume

./.venv/bin/python3 google_tv_skill.py pause ./.venv/bin/python3 google_tv_skill.py resume

Behavior and logging

• Default logging level is INFO. Use --verbose to enable DEBUG logging.
• Exit codes: 0 success, 2 adb/connect error, 3 resolution error (e.g., could not resolve YouTube ID), 4 adb command failure.

Dependencies & venv

• requirements.txt includes required Python packages. Install them into the skill venv at ./.venv.
• The script expects adb to be on PATH and yt-api to be available on PATH or at /Users/anthony/go/bin/yt-api.

Caching and non-destructive defaults

• The script stores the last successful device (ip and port) in .last_device.json in the skill folder.
• It will not attempt port scanning; this keeps behavior predictable and avoids conflicts with Google's ADB port rotation.

Troubleshooting

• If adb connect fails, run adb connect IP:PORT manually from your host to verify the current port.
• If adb connect is refused and you're running interactively, the script will prompt you for a new port and update .last_device.json on success.
• If yt-api resolution fails, ensure yt-api is installed or accessible at /Users/anthony/go/bin/yt-api.

Implementation notes

• The skill CLI code lives in google_tv_skill.py in this folder. It uses subprocess calls to adb and to yt-api when needed.
• For Tubi URL discovery, the assistant can use web_search to find canonical Tubi pages and pass the https URL to the skill.