Skill UI

This comprehensive benchmark suite evaluates the performance of Vision-Language Models (VLM) on complex video anomaly detection tasks using the SmartHome-Bench dataset. It covers a wide range of real-world smart home scenarios, including security breaches, elderly fall detection, and unusual pet behavior. Unlike single-frame analysis, this benchmark requires deep multi-frame video understanding, making it a robust tool for testing the advanced AI capabilities of video monitoring systems.

This skill guides the creation of professional, high-quality video content using a blend of programmatic frameworks and advanced AI generation models. It covers workflows for generating everything from data-driven product demos (using tools like Remotion and Hyperframes) to complex, original B-roll footage (using Veo, Runway, or Kling). Ideal for marketing teams needing efficient pipelines for explainers, social media clips, and ad creatives.

A comprehensive audio generation skill designed for professional multimedia content creation. It handles three types of audio output: music (jingles, beds), voiceovers (TTS), and sound effects (SFX). The skill intelligently routes requests to advanced models like Suno V5, Udio, MiniMax TTS, and ElevenLabs, producing a single, high-quality MP3 or WAV file ready for use in marketing, video, or podcast production.