ChatGPT Image Generation — SKILL.md

---
name: "ChatGPT Image Generation"
description: "Generate and edit images on ChatGPT via browser automation — create images from text, use style presets, edit existing images. For browser automation of chatgpt.com/images."
version: "1.0.0"
author: "skynet"
category: "content"
agents: ["claude-code"]
tags: ["chatgpt", "images", "ai-art", "browser-automation", "generation"]
tools_required: ["chrome-mcp"]
---

# ChatGPT Image Generation

# ChatGPT Image Generation

Browser automation skill for chatgpt.com/images — ChatGPT's dedicated image generation interface.

## Navigation

Navigate to `chatgpt.com/images` to access the image generation interface.

## Core Interface Elements

### Chat Input
- The chat textbox accepts image generation prompts
- **IMPORTANT**: This is a `contentEditable` div — do NOT use `form_input`. Instead use `click` on the input area then `type` your prompt
- Click the **Send prompt** button to submit

### Style Suggestion Buttons
Pre-built style prompts appear as clickable buttons:
- Me as an emperor
- Reimagine my pet
- Give them a bowl cut
- Redecorate my room
- Turn into a keychain
- Style me
- Create a coloring page
- Create a cartoon
- Create a professional product photo

### File Upload
- **Add files and more** button — upload reference images for editing or style transfer
- Supports drag and drop as well

### Image Gallery
- Previous generations appear in a gallery view
- Generated images appear inline in the chat
- Images can be downloaded directly

### Page Navigation
- Navigation buttons to browse through generations

## Tips
- Be descriptive in prompts for better results
- Upload reference images for editing tasks
- Style presets are good starting points — customize from there
- Generated images can be iterated on in the same conversation

## No Project System Warning

ChatGPT Images has NO project system or persistent memory. Agents must include full context in every prompt:
- Who the user is and their account details
- What they are working on and why
- The complete goal and any relevant history
- All necessary URLs, credentials references, and prior results

Never assume the agent remembers anything from previous interactions.

## Running on Mac Minis

These browser automation tasks run on Mac Minis via mac-control on port 8200:
- **Vault**: 192.168.86.27:8200
- **Bots**: 192.168.86.50:8200
- **Jarvis**: 192.168.86.51:8200

Use the mac-control API to send browser automation commands. Example:
```
POST http://192.168.86.27:8200/execute
```