# Create AI Videos with inference.sh

Choosing the right video model and CLI input can slow production. This skill gives Claude, Codex, and Claude Code clear inference.sh video workflows.

## Install

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

## Metadata

- Status: approved
- Slug: inference-sh-skills-ai-video-generation
- Skillstore revision: r1
- Version status: missing
- Tree hash: c761a31f25d863b6eaec7d754ee723e340c7344114abf26774288c4c8ce2e663
- Author: inference-sh-skills
- GitHub username: inference-sh-skills
- License: MIT
- Repository: https://github.com/inference-sh-skills/skills/tree/main/tools/video/ai-video-generation/
- Ref: a25199bc7d6b82598536822d1738eb5d5f54025b
- 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: 75
- Public page: https://skillstore.pages.dev/skills/inference-sh-skills-ai-video-generation
- Manifest: https://skillstore.pages.dev/api/skills/inference-sh-skills-ai-video-generation/manifest

## Capabilities

- Recommends model IDs for text-to-video, image-to-video, reference-to-video, lipsync, editing, upscaling, foley, and merging.
- Shows belt CLI examples for Google Veo, Seedance, Wan, Grok, OmniHuman, Fabric, and other video apps.
- Supports browsing available video apps with the video category command.
- Covers input fields such as prompt, duration, resolution, image URL, audio URL, and video URL.
- Links related skills for avatars, narration, image generation, and social posting.

## Use Cases

- Create Social Clips: Generate short videos from prompts or images for posts, campaigns, and tests.
- Prototype Product Videos: Select model examples for explainer clips, product motion, or background changes.
- Build Video Pipelines: Combine generation, upscaling, foley, and clip merging through repeatable CLI steps.

## Prompt Templates

### Make a Short Prompt Video

```
Use ai-video-generation to create a five second text-to-video clip about [scene]. Recommend the best fast model and ask before running.
```

### Animate an Image

```
Use ai-video-generation to turn [image URL] into a short video. Include motion guidance, duration, and a safe model choice.
```

### Create a Talking Avatar

```
Use ai-video-generation with [portrait URL] and [audio URL] to prepare a lipsync or avatar workflow. Explain privacy checks first.
```

### Plan a Multi-Step Video Pipeline

```
Use ai-video-generation to plan a pipeline for [goal] with generation, upscaling, foley, and merging. Compare model choices before any command.
```

## Limitations

- Requires the inference.sh belt CLI and an authenticated account.
- Video jobs may use external services and account credits.
- Sample media URLs are placeholders and must be replaced.
- The skill does not list current prices, quotas, or provider content rules.

## Best Practices

- Confirm media rights before using portraits, voices, product footage, or branded assets.
- Start with fast or economical models before requesting high resolution output.
- Review prompts, durations, and URLs before running jobs that may use credits.

## Anti Patterns

- Do not run generation jobs with private media unless external processing is approved.
- Do not leave placeholder URLs in requests.
- Do not request maximum duration or resolution before testing the prompt.

## Security Audit

- Audited at: 2026-07-05T17:44:19.797\+00:00
- Summary: The static Ruby backtick and hardcoded URL findings are false positives caused by Markdown code spans, fenced command examples, model IDs, and documentation links. No prompt injection or hidden malicious behavior was found in SKILL.md. The remaining concern is expected external processing of prompts and media URLs through the inference.sh belt CLI.

## Stats

- Views: 81
- Downloads: 8
- Favorites: 0
- Popularity score: 0