# Generate Images with Gemini Web

Teams need quick Gemini text and image generation from local prompts. This skill runs Gemini Web through Chrome cookies and saves generated images.

## Install

```bash
npx skillstore add jimliu/baoyu-gemini-web
```

## Metadata

- Status: approved
- Slug: jimliu-baoyu-gemini-web
- Skillstore revision: r1
- Version status: missing
- Tree hash: 609ce2895d2d1d9a3bdc2295fbfd5b44edf0a9e2396c06ad6e913968b0ada17e
- Author: JimLiu
- GitHub username: JimLiu
- License: MIT
- Repository: https://github.com/JimLiu/baoyu-skills/tree/main/skills/baoyu-gemini-web
- Ref: a06681402992ceae98ba04d54cfd4ab004862696
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: blocked
- Manual install advisory: allowed\_with\_warning
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, network, filesystem, env\_access
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/jimliu-baoyu-gemini-web
- Manifest: https://skillstore.pages.dev/api/skills/jimliu-baoyu-gemini-web/manifest

## Capabilities

- Runs text prompts through Gemini Web and returns plain text responses.
- Generates images from prompts and saves the first available image to disk.
- Reads prompt text from command arguments, standard input, or prompt files.
- Authenticates by opening Chrome and caching Gemini cookies for later runs.
- Supports Gemini web model selection with fallback to a flash model.
- Can emit structured command output for downstream automation.

## Use Cases

- Create campaign images: Generate draft visuals from concise creative prompts and save them for review.
- Automate Gemini drafts: Run repeatable Gemini text prompts from scripts or prompt files during content workflows.
- Prepare concept variations: Produce fast visual options from aspect ratio and style instructions before design refinement.

## Prompt Templates

### Basic text response

```
Use Gemini Web to answer this question in three direct paragraphs: [question].
```

### Generate an image

```
Generate an image of [subject] with [style], [lighting], and [aspect ratio]. Save it to [file name].
```

### Use prompt files

```
Read the system guidance from [file one] and the creative brief from [file two], then generate the requested output.
```

### Advanced image iteration

```
Use the attached image as the base. Change [specific elements], preserve [must-keep details], and output a polished final image.
```

## Limitations

- Requires access to Gemini Web through a Google account and a local Chrome-compatible browser.
- Stores sensitive Gemini cookies locally or accepts them through environment variables.
- Sends prompts, selected files, and image requests to Google services.
- Relies on undocumented Gemini Web endpoints that may change without notice.

## Best Practices

- Use a dedicated Google account with limited data exposure for Gemini Web automation.
- Review prompt files and attachments before sending them to Gemini.
- Store cookie files in a private directory with restrictive permissions.

## Anti Patterns

- Do not use a primary personal Google account for unreviewed automation.
- Do not send confidential files unless your policy allows Google processing.
- Do not depend on Gemini Web endpoints for production-critical workflows without fallback plans.

## Security Audit

- Audited at: 2026-07-06T15:37:57.665\+00:00
- Summary: The skill performs the advertised Gemini Web automation, but it captures and stores Google authentication cookies, launches Chrome, and sends prompts or files to Google endpoints. Most backtick detections are false positives from TypeScript templates or Markdown fences, while network, cookie, filesystem, and child-process risks remain confirmed. No prompt-injection attempt was found in the reviewed files.

## Stats

- Views: 168
- Downloads: 14
- Favorites: 0
- Popularity score: 0