Opencv

This is the standard library for computer vision work. You'll reach for it whenever you need to read images or video, detect edges and contours, track objects, or run face detection. It's built in optimized C++ so it's fast even from Python, with over 2500 algorithms ready to use. One critical gotcha: OpenCV uses BGR instead of RGB by default, so you'll need to convert colors when working with matplotlib or other libraries. The skill documentation covers the essentials well, from basic image I/O patterns to video processing loops, plus anti-patterns like using Python loops over pixels instead of vectorized operations. Works great for real-time applications where performance matters.

Install to Claude Code

npx -y skills add tondevrel/scientific-agent-skills --skill opencv --agent claude-code