# Generate AI Videos with RunComfy

Choosing video models and input settings can slow creative work. This skill maps your request to RunComfy routes with model-specific prompts and CLI examples.

## Install

```bash
npx skillstore add gencraft-labs/ai-video-generation
```

## Metadata

- Status: approved
- Slug: gencraft-labs-ai-video-generation
- Skillstore revision: r1
- Version status: missing
- Tree hash: 98fbb84b4eb09f5a24e429bfa7efb02d6d954259be9e45a4899423f83dc30453
- Author: gencraft-labs
- GitHub username: gencraft-labs
- License: MIT
- Repository: https://github.com/gencraft-labs/skills/tree/5d625f59c710d7a33316b50b302b7158cdee5971/ai-video-generation
- Ref: 131ee0146d7c649f59d7a3b5d7f6118e296ae8cb
- 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, filesystem
- Quality score: 69
- Public page: https://skillstore.pages.dev/skills/gencraft-labs-ai-video-generation
- Manifest: https://skillstore.pages.dev/api/skills/gencraft-labs-ai-video-generation/manifest

## Capabilities

- Select documented text-to-video, image-to-video, and Veo extension routes based on creative intent.
- Prepare RunComfy CLI generation requests with model identifiers, JSON inputs, and a local output directory.
- Provide prompt guidance for motion, camera direction, reference roles, and inline audio descriptions.
- Configure Wan lip-sync examples using a user-provided audio URL.
- Compare Kling quality tiers and identify faster model variants for iteration.
- Explain documented CLI exit codes for input errors, authentication failures, and retryable conditions.

## Use Cases

- Create Short Social Videos: Choose a vertical image-animation route and describe motion, camera behavior, duration, and audio.
- Animate Product Photography: Prepare a Veo image-to-video request for a controlled product rotation using an approved product image URL.
- Plan Cinematic Campaign Clips: Compare Seedance and Kling routes for cinematic framing, reference conditioning, character sequences, and quality tiers.

## Prompt Templates

### Beginner: Draft a Video Request

```
Prepare a five-second text-to-video request for a paper boat drifting across a pond. Recommend a documented model and prompt. Do not submit yet.
```

### Intermediate: Animate a Product Still

```
Use my approved image URL [image URL] for a product rotation. Keep the camera fixed. Prepare a Veo request for review before submission.
```

### Advanced: Match a Voiceover

```
Prepare a Wan lip-sync request using my approved audio URL [audio URL]. Describe a studio presenter. Confirm current input requirements before submission.
```

### Expert: Compare Production Routes

```
Plan a cinematic campaign clip using my approved references [references]. Compare Seedance and Kling tiers. Check current schemas and costs. Await approval before generation.
```

## Limitations

- Requires the external RunComfy CLI, authentication, network access, and any applicable service charges.
- Contains instructions only; model availability, rankings, output quality, and CLI security claims were not independently verified.
- Inputs and duration limits vary by model; detailed extension workflows require the linked documentation or a separate skill.
- Generation uses a remote service; arbitrary prompt text still requires safe shell argument handling.

## Best Practices

- Check the current model schema, service charges, and desired duration before approving a generation job.
- Use only authorized reference assets and task-specific URLs; keep API tokens out of prompts and logs.
- Pass prompt data as safely encoded arguments; do not assume JSON encoding alone makes shell interpolation safe.

## Anti Patterns

- Treating static ranking claims or exact motion examples as guaranteed generation results.
- Pasting untrusted prompt text into shell commands without correct argument quoting.
- Submitting confidential references or starting paid jobs without user authorization.

## Security Audit

- Audited at: 2026-11-14T14:43:05.066\+00:00
- Summary: All 154 static detections are false positives involving Markdown formatting, reference links, documented token storage, or model-selection advice. One semantic risk concerns an unconditional shell-safety claim without guidance for safely quoting arbitrary prompts. No evidence found of malicious intent, credential exfiltration, or instructions to override the audit; the external CLI implementation was not provided.

## Stats

- Views: 0
- Downloads: 1
- Favorites: 0
- Popularity score: 0
