# Create and Edit Images with Qwen-Image-2

Creating polished images and text-heavy graphics often requires several tools. This skill runs Qwen-Image-2 models through inference.sh for generation, editing, and controlled output.

## Install

```bash
npx skillstore add inferen-sh/qwen-image-2
```

## Metadata

- Status: approved
- Slug: inferen-sh-qwen-image-2
- Skillstore revision: r2
- Version status: missing
- Tree hash: 8055502e7e6d8b5bd4f2163c390126b70098a57536322543526fd4f12bd4fbfc
- Author: inferen-sh
- GitHub username: inferen-sh
- License: MIT
- Repository: https://github.com/inferen-sh/skills/tree/main/tools/image/qwen-image-2/
- Ref: 4121de961d1b6f2ffca856260e239505c302452c
- 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: 69
- Public page: https://skillstore.pages.dev/skills/inferen-sh-qwen-image-2
- Manifest: https://skillstore.pages.dev/api/skills/inferen-sh-qwen-image-2/manifest

## Capabilities

- Generates images from text prompts with Qwen-Image-2.0.
- Uses Qwen-Image-2.0-Pro for detailed text rendering and posters.
- Edits images using one to three reference-image URLs.
- Requests one to six output images in a single run.
- Controls width, height, negative prompts, prompt extension, watermarking, and seed.
- Provides model search, sample input, and execution workflows for the belt CLI.

## Use Cases

- Create Campaign Graphics: Generate posters with specified headlines, colors, layouts, and dimensions using the Pro model.
- Produce Concept Variations: Generate several visual directions from one prompt for rapid comparison and refinement.
- Edit Reference Images: Combine subject, clothing, and pose references into a new image through multi-reference editing.

## Prompt Templates

### Generate a Basic Image

```
Create an image of [subject] in [style] with [lighting]. Use Qwen-Image-2.0 and return one image.
```

### Create Several Concepts

```
Generate [number] variations of [concept]. Use [width] by [height] pixels and avoid [unwanted details].
```

### Design a Text-Heavy Poster

```
Use Qwen-Image-2.0-Pro to design a [format] poster. Include the exact text "[copy]" with [typography and placement].
```

### Combine Multiple References

```
Edit these authorized reference images: [URLs]. Preserve [elements], combine [attributes], use seed [seed], and describe the final composition before running.
```

## Limitations

- Requires the inference.sh belt CLI, authentication, and network access.
- Prompts and reference-image URLs are processed by an external service.
- Image dimensions must remain within the documented pixel limits.
- Generated text and visual details may still require review or regeneration.

## Best Practices

- Use the Pro model for posters, labels, and other text-heavy designs.
- Place required visible text in quotes and specify its color, style, and position.
- Use a fixed seed when comparing prompt changes or reproducing an approved result.

## Anti Patterns

- Do not submit confidential images or private URLs without authorization.
- Do not request dimensions outside the documented limits.
- Do not rely on generated text without checking spelling, layout, and legibility.

## Security Audit

- Audited at: 2026-08-06T11:04:51.449\+00:00
- Summary: Most static alerts are false positives caused by Markdown backticks, fixed CLI examples, public links, and an image parameter name. The unpinned npx skill installations are genuine supply-chain risks, and reference-image editing sends user data to an external processor.

## Stats

- Views: 145
- Downloads: 10
- Favorites: 0
- Popularity score: 0