What problem does it solve? Creating AI-generated videos requires navigating multiple model families, authentication modes, parameter constraints, and long-running async tasks. This Skill guides an agent through the full ListenHub video generation workflow—auth detection, parameter collection, cost estimation, submission, polling, and result download—without manual CLI guesswork. ## Core Features & Use Cases - Three model families: HappyHorse (text-to-video, image-to-video, reference-image, video-edit), SeeDance (frame mode with first/last frames, reference video/audio), and PixVerse (nine capabilities including lip sync, mimic, restyle, fusion, transitions, and a marketing ad agent). - Guided interaction flow: Collects prompts, reference assets, resolution, aspect ratio, and duration one question at a time, then shows a cost estimate and requires explicit confirmation before submitting. - Dual auth support: Detects internal OAuth or OpenAPI key authentication and routes commands accordingly, with PixVerse restricted to OpenAPI mode. - Use Case: Ask the agent to lip-sync an existing video to a TTS voiceover; it routes to the PixVerse lip_sync capability, collects the source video ID and speaker text, submits the task, polls until completion, and saves the MP4 to your working directory. ## Quick Start Ask the agent to generate a 5-second 1080p video of a cyberpunk city night scene using the video-gen skill.
