video-tutorial

Extracts code, commands, and steps from video tutorials using Gemini's visual analysis.

Updated Oct 7, 2022
One-click install
npx skills add https://github.com/tamagusko/linux-cfg --skill video-tutorial-tamagusko
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: video-tutorial
Source: https://github.com/tamagusko/linux-cfg/tree/main/dotfiles/claude/skills/video-tutorial
Command: npx skills add https://github.com/tamagusko/linux-cfg --skill video-tutorial-tamagusko

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Video tutorials hide their most important information on screen — the code being typed, the commands run, the versions in title bars — which transcripts and captions completely miss. This Skill watches the video's visual track through Gemini and turns it into a structured, replicable guide you can actually learn from and build against. ## Core Features & Use Cases - Visual-track extraction: Sends a YouTube URL or local video file to Gemini and recovers exact code, commands, versions, and UI actions shown on screen, with timestamps and illegible sections explicitly flagged. - Critical review and replication: Reads the extraction skeptically, verifies APIs and versions against real documentation, then actually rebuilds the project using locally installed versions and reports what worked and what failed. - Improvement over the source: Uses the extraction's Weaknesses section to fix outdated, insecure, or fragile tutorial patterns, clearly documenting every departure from the video. - Use Case: You find a 40-minute YouTube screencast building an auth middleware. Run the Skill to get the exact code and steps from the screen, learn why the approach works, reproduce it with your installed dependency versions, and receive a hardened version with the tutorial's shortcuts fixed. ## Quick Start Analyze this YouTube tutorial at the given URL, explain the technique it teaches, and rebuild the project it demonstrates using my current environment.

Frequently Asked Questions about video-tutorial

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I extract code shown in a YouTube tutorial video?

Run the analyze_video.py script with the YouTube URL and a GEMINI_API_KEY set. Gemini watches the visual track and transcribes on-screen code, commands, and versions into structured Markdown with timestamps, marking unreadable sections as ILLEGIBLE rather than guessing.

Can Gemini analyze a local video file instead of a YouTube link?

Yes, pass a local file path instead of a URL. The script uploads it through the Gemini Files API using resumable upload, polls until processing completes, then runs the same extraction prompt against the uploaded file.

Why use video analysis instead of a transcript for tutorials?

Transcripts only capture narration, while tutorials keep critical information on screen: exact code, terminal commands, menu clicks, and version numbers. Visual-track analysis recovers those details that a spoken-word transcript permanently loses.

What are the limitations of extracting code from video with Gemini?

Blurry or truncated screens produce ILLEGIBLE markers that require manual reconstruction, and inferred versions may not match reality. Long videos also consume large token counts in a single billed API call, so cost scales with video length.

Why does the Gemini API return a 404 for a listed model?

The models endpoint lists ids that are not actually callable for every account, such as gemini-2.5-pro for new users. Verify a model responds before relying on it, since appearing in the listing is not proof it works.