Gemini Watch 👁️

Gemini Watch is a native, ultra-lightweight extension for the Gemini CLI that gives the agent "eyes" to watch YouTube videos.

Instead of relying on heavy MCP servers or complex setups, gemini-watch provides a simple, zero-friction script that allows the CLI to instantly fetch video transcripts, timestamps, and metadata so it can summarize, analyze, and discuss YouTube content with you.

✨ Features

• Zero-Friction: No API keys required. No complex server configuration.
• Timestamps: Automatically formats transcripts with [MM:SS] timestamps so the agent can point you to the exact moment a topic was discussed.
• Metadata Extraction: Natively extracts the Video Title and Channel Name to provide immediate context before analyzing the transcript.
• Auto-Installing Dependencies: The underlying script automatically installs the necessary Python libraries (youtube-transcript-api) on its first run if they are missing.
• Native Feel: Designed purely via the GEMINI.md instruction model, making it incredibly fast and integrated natively into the agent's workflow.

🛠️ Prerequisites

To use this extension, you must have Python installed on your system, as the transcript engine is powered by a Python script.

1. Ensure Python 3 is installed and accessible from your terminal (python or py on Windows).
2. The script will automatically install youtube-transcript-api via pip, but you can install it manually if you prefer:

   pip install -r requirements.txt

📦 Installation

You can install this extension directly into Gemini CLI using the following command (replace <your-username> with your actual GitHub username once published):

gemini extensions install https://github.com/sudoax0n/gemini-watch

Note: If you have cloned this repository locally, you can link it instead:

gemini extensions link ./gemini-watch

🚀 Usage

Once installed, simply drop a YouTube link into your Gemini CLI session and ask it a question!

Example Prompts:

• "Summarize this video: https://www.youtube.com/watch?v=dQw4w9WgXcQ"
• "Extract the key takeaways from this lecture: [URL]"
• "At what timestamp did they talk about 'React 19' in this video? [URL]"

🧠 How it Works

1. When you provide a YouTube URL, the Gemini CLI agent reads the GEMINI.md instructions.
2. It extracts the Video ID from your URL.
3. It executes the scripts/get_transcript.py tool.
4. The script fetches the Title, Channel, and Timestamped Transcript.
5. The agent uses its massive context window to synthesize and answer your query.

🤝 Socials

Follow me for more Gemini CLI tools and hacks:

• X (Twitter): @beyondwudan
• GitHub: @sudoax0n

License

MIT License