Byted Seedance Video Generate

This one hooks into ByteDance's Doubao Seedance models (1.0 and 1.5) for generating videos from text prompts, images, or multimodal references. You can do text-to-video, guide with first and last frames for transitions, or throw in reference images, videos, and audio for style matching. The 1.5 model adds audio generation. It handles aspect ratios from 16:9 to portrait modes, outputs up to 1080p, and supports 2-12 second clips depending on the model. The async implementation polls tasks with configurable timeouts, returns video URLs, and includes fallback logic for model errors. Practical if you're building content tools that need programmatic video generation without training your own models.

Install to Claude Code

npx -y skills add bytedance/agentkit-samples --skill byted-seedance-video-generate --agent claude-code