# Optimize and Edit Images with MediaKit

Image workflows often need several tools for resizing, quality, editing, understanding, and privacy. This skill routes requests to the appropriate MediaKit image command and reference.

## Install

```bash
npx skillstore add volcengine/byted-mediakit-image
```

## Metadata

- Status: approved
- Slug: volcengine-byted-mediakit-image
- Version: 0.2.1
- Author version: 0.2.1
- Skillstore revision: r1
- Version status: valid
- Tree hash: 173e2cd8bc450f5673a67b4211a8be5ffcb0ed7eb7afc1e588d9b842dd94c8a5
- Author: volcengine
- GitHub username: volcengine
- License: MIT
- Repository: https://github.com/volcengine/mediakit-cli/tree/120059a6c4a089fb580edd24d8ecf93d687e3c96/skills/byted-mediakit-image
- Ref: 5526951beaeca22519572a471618555839b53b05
- 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: filesystem, external\_commands
- Quality score: 82
- Quality tier: silver
- Public page: https://skillstore.pages.dev/skills/volcengine-byted-mediakit-image
- Manifest: https://skillstore.pages.dev/api/skills/volcengine-byted-mediakit-image/manifest

## Capabilities

- Route image requests to 21 documented MediaKit image tools.
- Resize, compress, slim, crop, rotate, flip, and round image corners.
- Adjust color, sharpen, enhance, blur, mosaic, watermark, and invert images.
- Remove backgrounds, erase common text or icons, and detect faces for privacy blur.
- Read image text with OCR and inspect metadata, average hue, alpha, and blurhash.
- Evaluate image quality and perform face-aware smart cropping.

## Use Cases

- Prepare Product Assets: Resize, compress, crop, sharpen, and watermark product images for storefronts and marketing channels.
- Protect Sensitive Images: Blur faces, mosaic private regions, remove backgrounds, and erase visible text from approved media workflows.
- Review Visual Datasets: Extract image text, inspect metadata, score quality, and create consistent crops for content and training datasets.

## Prompt Templates

### Resize an Image

```
Resize this image to 1200 pixels wide while preserving its aspect ratio. Return the processed file.
```

### Compress for Web Delivery

```
Compress these images for web delivery with quality priority. Preserve their formats unless conversion is necessary.
```

### Create a Privacy-Safe Set

```
Detect and blur every face, then mosaic license plates in these images. Review the results for missed sensitive regions.
```

### Build a Dataset Pipeline

```
For this image batch, extract OCR text, inspect metadata, score quality, and smart-crop each image to 4:3. Report exceptions.
```

## Limitations

- The skill requires the mediakit-cli binary and its required authentication environment.
- Cloud tools require real input resources and may depend on service availability.
- OCR supports printed Simplified Chinese and English, with recognition limits on difficult images.
- Image edits can reduce detail, and privacy masking should be reviewed before publication.

## Best Practices

- State the intended output, dimensions, quality preference, and format before processing.
- Use the image-size guide when compression requirements are ambiguous.
- Review OCR, privacy masking, and automated crop results before distribution.

## Anti Patterns

- Do not request compression without stating whether quality or exact size matters.
- Do not assume masking removes every sensitive detail without reviewing the output.
- Do not provide invented input files, credentials, or optional parameters.

## Security Audit

- Audited at: 2026-09-12T11:54:31.357\+00:00
- Summary: All 53 static findings are false positives. The flagged path sequences are documentation links, the backtick matches are Markdown formatting around documented commands, and entropy heuristics reflect localized Markdown text; no prompt injection or malicious payload evidence was found.

## Stats

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