YouTube Guitar Tab Parser converts a YouTube guitar lesson video into a PDF containing the guitar tablature. The tool requires Node.js version 20 or later, yt-dlp, ffmpeg, and an Anthropic API key. After installing dependencies and building the project, the user provides a video URL. The program downloads the video up to a specified resolution, extracts frames at regular intervals, and uses Claude vision to identify the region containing sheet music. Each frame is cropped to that region. A perceptual hash filters near-identical consecutive crops to reduce vision API calls. Claude reads the bar number printed at the start of each line and determines whether the crop contains sheet music. Only one cropped image is retained per distinct bar number, and non-tab images are excluded. The distinct tab lines are stacked vertically into a PDF document, with the video title used as the file name and as metadata. The PDF is saved to the out/.pdf directory. Progress messages appear on standard error, and the output path is printed to standard output. Command-line options allow adjustment of screenshot interval, vision model, sample count, deduplication threshold, maximum download resolution, and retention of temporary files.