# Generate LibTV Images and Videos

Creative teams need fast image and video generation without manual job tracking. This skill sends prompts to LibTV, polls progress, uploads references, and downloads results.

## Install

```bash
npx skillstore add libtv-labs/libtv-skill
```

## Metadata

- Status: approved
- Slug: libtv-labs-libtv-skill
- Skillstore revision: r1
- Version status: missing
- Tree hash: ab43aa39dfc27d01d17b566ef82fb1046fc18d61e97a754b41d1d49c828c2621
- Author: libtv-labs
- GitHub username: libtv-labs
- License: MIT
- Repository: https://github.com/libtv-labs/libtv-skills/tree/main/skills/libtv-skill
- Ref: 62e2a730c5cd74eab4c7164309d810de660fcea3
- 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: network, env\_access, filesystem, external\_commands
- Quality score: 50
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/libtv-labs-libtv-skill
- Manifest: https://skillstore.pages.dev/api/skills/libtv-labs-libtv-skill/manifest

## Capabilities

- Creates new LibTV agent-im sessions for image and video generation requests.
- Sends follow-up messages to existing LibTV sessions by session ID.
- Uploads user-selected image or video reference files to LibTV OSS.
- Polls session messages to find generated image and video result URLs.
- Downloads generated media from LibTV result URLs into local folders.
- Switches the LibTV project bound to the configured access key.

## Use Cases

- Create Campaign Visuals: Generate image concepts, short videos, or product visuals from a brief and collect the finished files locally.
- Edit Reference Media: Upload an existing image or video, request object changes or style transfer, and retrieve the edited result.
- Track Long Generation Jobs: Poll LibTV sessions for progress, continue a conversation, and share final result links when the job completes.

## Prompt Templates

### Start a Simple Image

```
Create a calm product hero image for a white ceramic mug on a clean desk. Save the result when it finishes.
```

### Generate a Short Video

```
Create a five-second vertical video of a city bicycle moving through light rain at night. Use a cinematic but realistic style.
```

### Edit Uploaded Media

```
Upload this reference video and change the paper boats into white paper hearts. Keep the original camera movement and lighting.
```

### Plan and Retrieve a Storyboard

```
Create a nine-panel 16:9 storyboard for the next scene of this story. Poll until complete and download all generated panels.
```

## Limitations

- Requires LIBTV\_ACCESS\_KEY and network access to LibTV services.
- Uploads local reference media to LibTV, so sensitive files need approval.
- Only supports image and video uploads under 200 MB.
- Generation quality and timing depend on LibTV backend models and queues.

## Best Practices

- Confirm before uploading private or client-owned media to LibTV.
- Keep session IDs and project IDs with the project notes for later polling.
- Use clear creative goals and let the LibTV backend handle prompt expansion.

## Anti Patterns

- Do not upload confidential source files without explicit user approval.
- Do not rewrite the user prompt into a complex prompt unless asked.
- Do not expose project canvas links before the generation task completes.

## Security Audit

- Audited at: 2026-07-05T19:49:37.158\+00:00
- Summary: The skill intentionally contacts LibTV services, reads a bearer access key, uploads user-selected media, polls sessions, and writes downloaded results locally. Most static command, reconnaissance, and entropy findings are false positives from Markdown examples or LibTV project identifiers. The main risks are credential-bearing network calls, configurable API hosts, user-controlled command arguments, and local file writes.

## Stats

- Views: 123
- Downloads: 75
- Favorites: 1
- Popularity score: 0