What problem does it solve?
Creating AI-generated marketing videos and images across multiple models (Seedance 2.0, Sora 2, Veo 3.1, Kling 3.0, Grok Video, Nano Banana) requires knowing each model's API parameters, prompt formulas, credit costs, and polling behavior. This Skill handles the full Arcads external API workflow so you can produce ad creative without memorizing endpoint schemas or vendor prompting guides.
Core Features & Use Cases
- Multi-model video and image generation: Routes requests to the unified
POST /v2/videos/generate endpoint or image/b-roll/scene endpoints with the correct model value, duration, aspect ratio, and reference inputs.
- Prompt library and workflow guidance: Loads model-specific prompting formulas (UGC, product hero, premium reveal, character sheets, influencer recreation) before composing any prompt.
- Cost control and QA gates: Estimates credit costs from historical logs before generating, enforces a dialogue confirmation gate for talking videos, and runs visual QA with retries on generated still images.
- Use Case: Ask for a 15-second UGC-style product review video. The Skill creates a dated project folder in Arcads, drafts the prompt and dialogue for your approval, shows a credit estimate, generates the video via Seedance 2.0, polls until ready, and delivers watch URLs.
Quick Start
Ask the agent to generate a 15-second UGC-style product review video with Seedance 2.0 using your Arcads API key configured in the .env file.