What is GF(3) skill mapping?

It is a specialized triadic classification system used within the skill to ensure a balanced and mathematically consistent categorization of technical concepts across video segments.

What file formats does Sense support?

Sense supports common video containers such as .mp4 and .mkv, alongside standard subtitle formats like WebVTT (.vtt) and SubRip (.srt).

Do I need any external dependencies to run this?

Yes, Sense typically requires FFmpeg for frame extraction and a valid Mathpix API key to process mathematical OCR requests.

How does it handle mathematical equations in videos?

The skill utilizes the Mathpix API to perform high-accuracy OCR on extracted video frames, converting visual formulas and diagrams directly into LaTeX strings.

Where is the indexed lecture data stored?

All processed data, including timestamps, transcripts, and LaTeX equations, is stored in a local DuckDB database for efficient SQL-based querying.

Sense Video Lecture Indexer

Name: Sense Video Lecture Indexer
Author: plurigrid

byplurigrid

•

学習とドキュメンテーション

Extracts structured knowledge, LaTeX equations, and diagrammatic data from video lectures using subtitle alignment and OCR.

Sense is a specialized tool for Claude Code designed to transform video lectures into searchable, structured knowledge bases. It automates the process of parsing subtitles, detecting slide transitions via visual and audio cues, and extracting mathematical equations or diagrams through Mathpix OCR. By indexing this data into a DuckDB database, it allows users to perform complex queries across video content, map specific segments to technical skills, and maintain a high-fidelity record of educational or technical presentations, making it ideal for researchers and students working with dense technical media.

主な機能

01Searchable DuckDB indexing for fast querying of video transcripts and LaTeX content

02Intelligent slide transition detection based on silence gaps and visual changes

03Frame-accurate diagram and equation extraction using FFmpeg and Mathpix OCR

048 GitHub stars

05GF(3)-balanced skill mapping for categorizing content into domain-specific expertise

06Automated subtitle parsing and timestamp alignment for WebVTT and SRT formats

ユースケース

01Creating searchable archives of university or technical engineering lectures

02Extracting mathematical formulas and diagrams from video presentations into LaTeX format

03Building automated skill-mapping pipelines for educational content repositories

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add plurigrid/asi sense

For use in Claude.ai and ChatGPT

Download Skill