What problem does it solve? Accessing multiple hosted image and video generation models normally requires separate vendor accounts, API keys, and endpoint integrations. This Skill routes all media generation through a single Atlas Cloud gateway using one ATLASCLOUD_API_KEY, with preflight cost checks and consistent result provenance. ## Core Features & Use Cases - Multi-model video generation: Route text-to-video, image-to-video, and reference-to-video jobs through Seedance 2.5/2.0, Gemini Omni Flash, and MiniMax H3 with exact model IDs and per-second pricing. - Multi-model image generation and editing: Generate, edit, or decompose images with Seedream 5.0 Pro/Lite, GPT Image 2, and Nano Banana 2, supporting multi-image edit inputs. - Preflight cost and catalog validation: Read the live model catalog before every paid call, announce provider, model, request count, and estimated cost, and fail invalid parameters before billing. - Use Case: A content team needs a 10-second product video from reference images and audio. The Skill selects minimax/h3/reference-to-video, confirms the $0.10/s cost, submits the job, polls the prediction, and saves the output into the active project folder. ## Quick Start Ask the agent to generate a 10-second 2K video from your reference image and audio using MiniMax H3 through Atlas Cloud and save it into the current project folder.
