# Create AI Videos with 40\+ Models

Choosing and operating video generation models requires many separate workflows. This skill provides direct belt CLI guidance for generation, animation, editing, audio, and finishing tasks.

## Install

```bash
npx skillstore add 101-skills/ai-video-generation
```

## Metadata

- Status: approved
- Slug: 101-skills-ai-video-generation
- Skillstore revision: r1
- Version status: missing
- Tree hash: eed05eb86acf119032e90d7d43c271bad6002415deb8e2221c1d69e7e8c358b4
- Author: 101-skills
- GitHub username: 101-skills
- License: MIT
- Repository: https://github.com/101-skills/skills/tree/main/tools/video/ai-video-generation/
- Ref: d71c7417a35d5c2624161bd2fe8de8a41a362128
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: allowed
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 50
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/101-skills-ai-video-generation
- Manifest: https://skillstore.pages.dev/api/skills/101-skills-ai-video-generation/manifest

## Capabilities

- Selects documented models for text-to-video and image-to-video tasks.
- Runs reference-based character animation with supported Seedance and HappyHorse models.
- Creates talking-head and lipsync videos from portrait and audio inputs.
- Edits existing video through natural-language instructions.
- Upscales video, generates foley audio, and merges clips with documented utility apps.
- Browses available video applications through the belt CLI catalog.

## Use Cases

- Produce Social Video Concepts: Generate short visual concepts from campaign copy, then compare fast and high-quality model options.
- Animate Product Artwork: Turn approved product images into motion clips and add controlled camera movement or environmental effects.
- Build Training Presenters: Create consented talking-head clips from a portrait and recorded narration, then upscale the final video.

## Prompt Templates

### Generate a Simple Video

```
Create a short video of [subject] performing [action] in [setting]. Choose a fast text-to-video model and ask before running the command.
```

### Animate an Existing Image

```
Animate the image at [image URL]. Add [motion], preserve the main subject, and recommend a suitable image-to-video model before execution.
```

### Create a Consented Presenter

```
Create a talking-head video from [portrait URL] and [audio URL]. Confirm subject consent, explain external processing, then select a supported avatar model.
```

### Plan a Finished Video Pipeline

```
Design a pipeline for [goal] using generation, editing, foley, upscaling, and merging. Compare model tradeoffs and request approval before each external run.
```

## Limitations

- Requires the inference.sh belt CLI, an authenticated account, and network access.
- Media inputs must be available through URLs accepted by the selected provider.
- Supported parameters, duration, resolution, quality, and audio features vary by model.
- The skill does not verify media rights, subject consent, provider policies, or output accuracy.

## Best Practices

- Confirm the selected model, expected duration, resolution, and audio behavior before execution.
- Use media only when rights and subject consent are clear, and disclose external processing.
- Test a short, economical generation before requesting longer or higher-quality output.

## Anti Patterns

- Do not submit private portraits, recordings, or videos without informed consent and provider review.
- Do not assume every model accepts the same fields, durations, resolutions, or reference formats.
- Do not run installation, login, or paid generation commands without explicit user approval.

## Security Audit

- Audited at: 2026-07-12T09:56:19.126\+00:00
- Summary: The Ruby backtick alerts are Markdown formatting false positives, while one remote image reference can trigger a low-risk external request. Semantic review found broad belt command authorization and undisclosed third-party processing of user media. No prompt injection or malicious intent was found.

## Stats

- Views: 0
- Downloads: 6
- Favorites: 0
- Popularity score: 0