How can I reduce the token cost for video analysis?

You can reduce token usage by lowering the sampling frame rate (FPS) or setting the media resolution to 'low', which drops token cost per frame significantly.

Does this skill support private YouTube videos?

No, this skill currently only supports public YouTube videos due to API access limitations.

Can I limit the analysis to a specific part of a video?

Yes, you can configure video clipping by setting start and end offsets to analyze only a specific segment of the video.

What is the best way to handle large video files?

For files larger than 20MB or longer than one minute, it is recommended to use the Files API upload method for more reliable processing.

What video formats are compatible?

The skill supports major formats including MP4, MPEG, MOV, AVI, WMV, WEBM, and 3GPP.

Gemini Video Intelligence

Name: Gemini Video Intelligence
Author: jongwony

byjongwony

Data Science & ML

Analyzes and extracts deep insights from video files and YouTube URLs using the Google Gemini API.

About

The Gemini Video Intelligence skill empowers Claude to understand visual content deeply by leveraging the Google Gemini 3 Flash API. Whether working with local video files or public YouTube URLs, this skill allows users to generate comprehensive summaries, precise transcripts, and timestamped event logs. It is particularly useful for developers and researchers needing to automate video content analysis, answer specific questions about visual sequences, or optimize token usage through advanced controls like video clipping, frame rate adjustment, and resolution scaling.

Key Features

Direct integration with Gemini 3 Flash for high-speed multimodal analysis
Precise timestamping for scene changes and key events
Automated transcription and speaker-identified dialogue extraction
Advanced token optimization via clipping, frame rate, and resolution control
0 GitHub stars
Multi-source input support for local files and public YouTube URLs

Use Cases

Building automated video indexing systems with timestamped event logs
Performing natural language Q&A to locate specific information within video clips
Generating structured summaries and transcripts for long-form video content

About

Key Features

Direct integration with Gemini 3 Flash for high-speed multimodal analysis
Precise timestamping for scene changes and key events
Automated transcription and speaker-identified dialogue extraction
Advanced token optimization via clipping, frame rate, and resolution control
0 GitHub stars
Multi-source input support for local files and public YouTube URLs

Use Cases

Building automated video indexing systems with timestamped event logs
Performing natural language Q&A to locate specific information within video clips
Generating structured summaries and transcripts for long-form video content