# Generate AI Videos with CLI Models

AI video workflows require many model choices and exact CLI inputs. This skill maps common video goals to inference.sh commands, model families, and media preparation steps.

## Install

```bash
npx skillstore add inference-sh/ai-video-generation
```

## Metadata

- Status: approved
- Slug: inference-sh-ai-video-generation
- Skillstore revision: r1
- Version status: missing
- Tree hash: 2da958f42071a339370d73d1e3d3e3c3d4a5de1f0cf9bd9f77b272cf2e361325
- Author: inference-sh
- GitHub username: inference-sh
- License: MIT
- Repository: https://github.com/inference-sh/skills/tree/main/skills/ai-video-generation/
- Ref: a25199bc7d6b82598536822d1738eb5d5f54025b
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: blocked
- Manual install advisory: allowed\_with\_warning
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/inference-sh-ai-video-generation
- Manifest: https://skillstore.pages.dev/api/skills/inference-sh-ai-video-generation/manifest

## Capabilities

- Lists text-to-video model options, including Veo, Grok Video, and Seedance models.
- Provides image-to-video examples for Wan and Seedance workflows.
- Shows avatar and lipsync commands using portrait and audio URLs.
- Includes utility commands for video upscaling, foley sound, and clip merging.
- Shows how to browse video apps through the inference.sh CLI.

## Use Cases

- Create Campaign Video Drafts: Turn a marketing idea into a model choice, prompt direction, and generation plan.
- Animate Existing Visual Assets: Choose image-to-video or lipsync workflows for portraits, product shots, and short clips.
- Prototype Media Pipelines: Combine generation, foley, upscaling, and merging steps for repeatable video workflows.

## Prompt Templates

### Generate a Simple Clip

```
Create a short text-to-video plan for a calm product reveal. Recommend one model and describe the prompt inputs.
```

### Animate a Reference Image

```
Turn this image URL into an image-to-video plan. Choose a Wan or Seedance model and describe motion, duration, and aspect ratio.
```

### Create a Talking Avatar

```
Plan a talking-head video from my portrait URL and narration audio URL. Recommend Fabric, PixVerse, or OmniHuman.
```

### Build a Video Pipeline

```
Design a workflow that creates a product demo, adds foley sound, upscales the result, and merges final clips.
```

## Limitations

- Requires the inference.sh CLI, account access, and any needed provider permissions.
- Results depend on external model availability, pricing, quotas, and content policies.
- Examples require accessible media URLs for images, audio, or videos.
- Does not verify copyright, consent, or brand safety for generated media.

## Best Practices

- Use clear prompts with subject, motion, camera, setting, and duration details.
- Use only media that you own or have permission to process.
- Review external service costs, quotas, and content policies before running commands.

## Anti Patterns

- Do not process private portraits, voices, or unreleased product footage without approval.
- Do not run installer commands from marketplace content without independent verification.
- Do not choose the largest model before testing a low-cost draft workflow.

## Security Audit

- Audited at: 2026-07-05T17:21:38.304\+00:00
- Summary: The major confirmed issue is a curl-to-shell installer command in the quick start, which is unsafe for community marketplace content. The skill otherwise uses intended inference.sh CLI examples, but those commands send prompts and media URLs to external services.

## Stats

- Views: 299
- Downloads: 35
- Favorites: 0
- Popularity score: 0