Agent Skills
› siddsachar/row-bot
› video_guide
video_guide
GitHub指导AI通过文本生成短视频或让静态图片动态化。涵盖generate_video和animate_image工具的使用场景、提示词编写规范、默认参数设置及不同提供商（如Google Veo, xAI）的具体限制与错误处理策略。
Trigger Scenarios
用户请求生成视频或动画
用户要求将图片动起来
Install
npx skills add siddsachar/row-bot --skill video_guide -g -y
SKILL.md
Frontmatter
{
"icon": "🎬",
"name": "video_guide",
"tags": [],
"tools": [
"video_gen"
],
"description": "Guidance for generating videos and animating images.",
"display_name": "Video Guide"
}
- You can generate short video clips from text descriptions using generate_video, or animate a still image into a video using animate_image.
- Use generate_video when the user asks you to create, produce, or make a video, clip, animation, or motion content from a text description. Provide a detailed prompt including subject, action, style, camera motion, and ambiance. Short cinematic prompts work best.
- Use animate_image when the user wants to bring a still image to life — the last generated image, an attached/pasted image, or a file on disk. Describe how the image should move or animate.
- Video generation takes 1–6 minutes depending on the provider and settings. Let the user know it may take a moment. Do not make multiple video requests in parallel.
- If a provider returns a validation error such as INVALID_ARGUMENT, do not retry the same request unchanged. Surface the provider error once, then adjust parameters or switch providers.
- Default settings: 8 seconds, 16:9 aspect ratio, 720p resolution. Only change these if the user explicitly requests different values.
- Google Veo supports 4, 6, or 8 second durations. 1080p and 4k require 8 seconds. Aspect ratios: 16:9 or 9:16. Generates audio natively.
- For Google Veo in this release, do not infer or force person-generation settings from the prompt. Let the provider defaults apply.
- xAI supports 1–15 second durations. Aspect ratios: 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3. Resolutions: 480p or 720p.
- Generated videos are saved as MP4 files in the thread media folder and displayed inline in chat.
- Do NOT promise video editing, video extension, reference-image workflows, or fal-backed models — these are not available in this version.
Version History
- edc0ac0 Current 2026-07-24 22:30
