video▌

Given a YouTube video URL or ID, return title, channel, duration, full timestamped transcript segments, and whether captions are auto-generated or human-authored. Read-only.

Category: test

Custom video players with adaptive streaming, quality selection, and playback controls. \n \n Supports HLS adaptive bitrate streaming with automatic quality switching and error recovery, plus native Safari fallback \n Includes fullscreen, picture-in-picture, volume control, playback speed adjustment, and subtitle/caption tracks \n Provides quality selection UI, thumbnail preview on hover, and video upload with progress tracking \n Built with React hooks and hls.js library for cross-browser compa

./scripts/vtd.js can:

Forensic deconstruction of YouTube videos to extract viral formulas, hooks, and retention mechanics. \n \n Analyzes video transcripts across 11 systematic sections: hook architecture, structural blueprint, retention mechanics, emotional engineering, storytelling, linguistic patterns, algorithm signals, CTAs, and viral coefficient scoring \n Extracts reusable fill-in-the-blank script templates, power phrases, and transition patterns ready for immediate application to new content \n Includes imple

Unified audio generation dispatcher routing TTS, music, sound effects, and voice cloning to optimal models. \n \n Routes requests to ElevenLabs (TTS, voice cloning, SFX) or fal.ai (music) based on request type, with latencies ranging from <1s to ~15s \n Supports five audio capabilities: multilingual text-to-speech with voice selection, low-latency turbo TTS, background music composition, sound effect generation (up to 22 seconds), and voice cloning from audio samples \n Requires only ELEVEN

Unified access to 8 AI image generation models with automatic model selection and zero API key setup. \n \n Supports Midjourney, Flux (Pro/Dev/Schnell), Ideogram, Recraft, SDXL, and Nano Banana with automatic model routing based on user intent \n Handles Midjourney's async polling transparently; all models return consistent output format with image URLs \n Includes Midjourney actions (upscale, variation, reroll) and reference image support for style consistency \n All requests routed through hos

Category: test

You are an expert videographer and motion designer specializing in creating engaging promotional short-form videos for software projects. Your goal is to create a TikTok-style promotional video that showcases the project's key features and value proposition.

You are a creative strategist specializing in video ad deconstruction. Analyze video ads to extract hooks, persuasion tactics, targeting insights, and replicable patterns.