Skills gpt-image-2
📦

gpt-image-2

Low Risk ⚙️ External commands🌐 Network access📁 Filesystem access

Generate Images with GPT Image 2 on RunComfy

Also available from: doany-ai,agentspace-so

Create and edit images using OpenAI GPT Image 2 through the RunComfy cloud platform. This skill provides expert prompting patterns, size constraints, and model selection guidance so you get the best results on every generation.

Supports: Claude Codex Code(CC)
🥉 72 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "gpt-image-2". Generate a product photo of a ceramic water bottle on warm linen with the text 'AQUA+' on the label

Expected outcome:

High-resolution product image with accurate label text reading AQUA+, soft window lighting, subtle rim highlights, and neutral e-commerce background

Using "gpt-image-2". Edit this cafe photo: turn the background into a bright white studio sweep and add the headline 'OPEN STUDIO' in bold sans-serif, centered

Expected outcome:

Edited image with clean white-to-soft-gray studio background, centered bold OPEN STUDIO headline text, preserved subject identity and composition

Using "gpt-image-2". Create a Tokyo cafe storefront at dusk with the sign reading the Japanese characters for coffee in kana

Expected outcome:

Cinematic storefront image with warm interior glow, accurate Japanese text on wooden plaque, shallow depth of field, rule of thirds composition

Security Audit

Low Risk
v1 • 5/28/2026

Static analysis flagged 77 patterns across SKILL.md (58 shell backticks, 10 hardcoded URLs, 5 filesystem paths, 3 weak-crypto indicators, 1 reconnaissance). All findings evaluated as false positives: backticks contain markdown CLI documentation examples, URLs are legitimate RunComfy service endpoints, filesystem references document CLI config paths (~/.config/runcomfy/token.json), and crypto/reconnaissance patterns are misidentified text ("Exit codes" heading, jq pipe examples). No executable code, no prompt injection, no data exfiltration detected. Minor concerns: user prompts transmitted to RunComfy's external API, third-party CLI dependency, local token file storage.

1
Files scanned
212
Lines analyzed
6
findings
1
Total audits
Low Risk Issues (3)
Third-Party Service Data Transmission
User prompts and image URLs are transmitted to RunComfy's model API (model-api.runcomfy.net) for processing. This is the intended function but users should be aware their data leaves the local environment.
Third-Party CLI Dependency
Skill requires installation of @runcomfy/cli via npm. This introduces a supply chain dependency on an external package not controlled by the skill author.
Local Token Storage
RunComfy CLI stores authentication token in ~/.config/runcomfy/token.json. Documented as using mode 0600 permissions which provides adequate local protection.

Detected Patterns

Shell Backtick Execution (58 instances — All False Positives)Hardcoded URLs (10 instances — All False Positives)Filesystem Path Patterns (5 instances — All False Positives)
Audited by: claude

Quality Score

38
Architecture
100
Maintainability
87
Content
55
Community
84
Security
91
Spec Compliance

What You Can Build

E-Commerce Product Photography

Generate product images with accurate label text, brand-safe lighting, and consistent styling across product lines using GPT Image 2's precise text rendering.

Multilingual Brand Asset Creation

Create signage, posters, and packaging mockups with accurate text rendering in multiple languages from a single source asset using the edit endpoint.

Iterative Image Refinement

Edit images step by step, changing one attribute at a time while preserving composition, faces, and brand elements across multiple generations.

Try These Prompts

Simple Text-to-Image Generation
Generate an image of [describe subject] in [describe setting] with [describe mood or lighting] using GPT Image 2 on RunComfy.
Product Image with Embedded Text
Create a product photo of [product] on [surface or background], the label reads "[exact text]" in [font style], [lighting description], e-commerce ready, neutral background.
Image Edit with Preservation
Edit this image using GPT Image 2: [describe specific change]. Keep [list elements to preserve] unchanged. Use [size] for output.
Multi-Reference Image Composition
Compose a new image using GPT Image 2 edit: subject from image 1 and background from image 2. Match the lighting of image 2. Keep the pose and face identity from image 1 unchanged.

Best Practices

  • Quote all embedded text exactly as you want it to appear in the generated image for accurate text rendering
  • Change only one attribute per edit iteration such as lighting, background, pose, or text to maintain composition stability
  • Use compositional cues like rule of thirds, close-up, aerial view, or shallow depth of field directly in your prompts

Avoid

  • Do not combine conflicting instructions such as requesting no text while also specifying text on product labels
  • Avoid piling up multiple art styles like ukiyo-e plus watercolor plus cinematic in one prompt, which produces inconsistent results
  • Do not use extreme aspect ratios since only three fixed sizes are supported and unsupported ratios are auto-resized

Frequently Asked Questions

What is GPT Image 2 best at compared to other image models?
GPT Image 2 excels at embedded text rendering, logos, multilingual typography, and precise instruction following. It is ideal when accuracy matters more than artistic stylization.
Do I need an OpenAI API key to use this skill?
No. This skill uses RunComfy's hosted Model API. You only need a RunComfy account and the RunComfy CLI installed locally.
What image sizes are supported?
Three fixed sizes are supported: 1024x1024 for square, 1024x1536 for portrait, and 1536x1024 for landscape. The edit endpoint also supports auto to preserve the input aspect ratio.
How does the edit endpoint preserve elements of the original image?
Use keep statements in your prompt, such as keep the person's pose and face identity unchanged. The model holds composition stable when you change only one attribute at a time.
Is my prompt data sent to external servers?
Yes. Prompts are sent over HTTPS to RunComfy's model API at model-api.runcomfy.net. The CLI does not shell-expand prompts. Review RunComfy's privacy policy for details on data handling.
How do I cancel a running image generation?
Press Ctrl-C while the CLI is polling for results. This sends a cancel request to the API so you are not billed for unused GPU time.

Developer Details

File structure

📄 SKILL.md