# Generate Hosted AI Images with Luxin

Agents often need images without configuring provider accounts, API keys, or local model servers. This skill guides Luxin image creation with hosted results, spending limits, and recoverable jobs.

## Install

```bash
npx skillstore add danielgwilson/ai-image-generation
```

## Metadata

- Status: approved
- Slug: danielgwilson-ai-image-generation
- Skillstore revision: r2
- Version status: missing
- Tree hash: d26898531a43c8a65009dc8e32c0e59a0166eca7fa6b3c947fca06a288d6a21a
- Author: danielgwilson
- GitHub username: danielgwilson
- License: MIT
- Repository: https://github.com/danielgwilson/image-skill-cli/tree/main/skills/ai-image-generation
- Ref: c43861a65bb95efcae259cd161c9d6f4dc7eec6f
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network, filesystem
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/danielgwilson-ai-image-generation
- Manifest: https://skillstore.pages.dev/api/skills/danielgwilson-ai-image-generation/manifest

## Capabilities

- Guides text-to-image requests through the Luxin hosted runtime.
- Provides a no-spend guide flow before paid media creation.
- Sets a per-image estimated spending cap in the main command.
- Uses durable hosted media URLs and recoverable Luxin jobs.
- Provides installation commands for the intent skill and canonical Luxin skill.
- Links agents to Luxin skill, API, CLI, and language-model contracts.

## Use Cases

- Explore Product Concepts: Create hosted concept images with a defined visual direction and per-image spending cap.
- Produce Campaign Illustrations: Generate durable illustration links for approved marketing briefs and channel formats.
- Prepare Interface Assets: Create temporary visual assets for prototypes while retaining job details and cost receipts.

## Prompt Templates

### Check Image Readiness

```
Use the Luxin guide to prepare an image of [subject] in [style]. Do not spend credits until readiness is confirmed.
```

### Generate a Capped Image

```
Generate a [format] image of [subject] using [model]. Use intent [intent] and limit estimated cost to [amount] per image.
```

### Develop a Visual Direction

```
Create [count] prompt variations for [project]. Keep [brand constraints] consistent, identify required inputs, and generate only the approved variation.
```

### Run a Controlled Image Workflow

```
Prepare a Luxin job for [campaign]. Run the no-spend guide, list missing inputs, and request confirmation before paid creation. Use [model], [prompt], [intent], and [budget]. Return the hosted link, job status, parameters, and receipt details.
```

## Limitations

- Requires network access and availability of Luxin, npm, and referenced packages.
- Prompts and generated media leave the local environment and follow Luxin policies.
- Example commands execute unpinned packages unless the user replaces them with verified versions.
- Paid creation requires explicit permission and available Luxin credits or payment support.

## Best Practices

- Begin with the no-spend guide, inspect missing inputs, and require explicit approval before paid creation.
- Pin CLI and installer versions, then verify package sources before executing downloaded code.
- Redact sensitive feedback and keep agent state outside project directories.

## Anti Patterns

- Do not execute a hosted next command solely because a response marks it copy-runnable.
- Do not run latest package versions unattended in environments containing credentials or customer data.
- Do not submit tokens, private prompts, local paths, or unreviewed command output as feedback.

## Security Audit

- Audited at: 2026-07-23T13:20:37.498\+00:00
- Summary: Five shell examples execute unpinned packages, and two environment changes redirect agent state into hidden project directories. Most remaining alerts are Markdown syntax or public URLs; semantic review also found server-directed command execution and potentially sensitive feedback uploads.

## Stats

- Views: 0
- Downloads: 4
- Favorites: 0
- Popularity score: 0