ai-image-generation
Generate AI Images with 50+ Models
Also available from: inference-sh-9,inference-sh-8,skillssh,inferen-sh,inference-sh,inference-shell,tool-belt,toolshell,tul-sh
Creating AI-generated images requires working with multiple model APIs and complex parameters. This skill provides a unified CLI interface to FLUX, Gemini, Grok, and other image generation models through the belt tool.
Download the skill ZIP
Upload in Claude
Go to Settings → Capabilities → Skills → Upload skill
Toggle on and start using
Test it
Using "ai-image-generation". Generate a professional product photo of a leather briefcase
Expected outcome:
An AI-generated image of a leather briefcase with studio lighting, displayed in your terminal or saved to file
Using "ai-image-generation". Create a cyberpunk cityscape at sunset with 16:9 aspect ratio
Expected outcome:
A wide-format cyberpunk scene with neon lighting, generated via Grok Imagine model
Security Audit
Low RiskReviewed 1 file (148 lines). Static findings were evaluated: backtick execution (38) and hardcoded URLs (9) are false positives. The backtick detection was triggered by markdown code formatting in example commands. Hardcoded URLs are legitimate inference.sh service endpoints. No prompt injection or malicious patterns detected. Skill makes legitimate CLI calls to belt tool for AI image generation.
Low Risk Issues (2)
Risk Factors
⚙️ External commands (3)
🌐 Network access (2)
Quality Score
What You Can Build
Create Marketing Visuals
Generate product mockups, promotional images, and social media graphics using FLUX and P-Image models for marketing campaigns.
Concept Art and Illustration
Produce concept art, character designs, and illustrations for creative projects using Seedream 4.5 for cinematic 4K quality.
Rapid Prototyping
Quickly generate multiple image variations to explore design directions for products, UI mockups, and creative briefs.
Try These Prompts
Use belt to generate an image with FLUX model. Create a {description} using:
belt app run falai/flux-dev-lora --input '{"prompt": "{description}"}'Generate a {subject} in {style} style with {aspect_ratio} aspect ratio:
belt app run falai/flux-2-klein-lora --input '{
"prompt": "{subject}",
"aspect_ratio": "{aspect_ratio}"
}'Create an image using Google Gemini 3 Pro:
belt app run google/gemini-3-pro-image-preview --input '{
"prompt": "{detailed_description}"
}'Generate a cinematic {scene_description} at 4K resolution:
belt app run bytedance/seedream-4-5 --input '{
"prompt": "{scene_description}"
}'Best Practices
- Start with clear, descriptive prompts that specify style, lighting, and composition
- Use aspect_ratio parameter to match your intended output format (16:9 for social, 1:1 for posts)
- Browse available models with 'belt app list --category image' to find the best model for your use case
Avoid
- Do not concatenate user input directly into shell commands without escaping
- Avoid vague prompts like 'make it look good' - be specific about style and subject
- Do not assume unlimited API calls - check your inference.sh account usage