Virtual Try On

Convert clothing images into professional e-commerce photos by virtually dressing AI models with up to four garment images for online retail use.

Security Scan

The skill's runtime instructions and manifest require an API key and post user images to an external service (api.ngmob.com), but the package metadata does not declare that credential or provide provenance (homepage/source/consistent owner), creating an incoherent and potentially risky footprint.

Purpose & Capability

The skill description and SKILL.md describe an API-based virtual try-on service (api.ngmob.com), which reasonably requires an API key. However, the registry metadata declares no required environment variables or primary credential while the manifest and SKILL.md explicitly use Authorization: Bearer {{API_KEY}} / $API_KEY. Additionally there is no homepage/source and owner IDs/authorship are inconsistent (manifest/_meta/registry show mismatched or placeholder values), reducing provenance and trust.

Instruction Scope

Instructions are scoped to sending user-provided clothing image URLs to https://api.ngmob.com and polling for results — this matches the declared purpose. However the instructions require an API_KEY (curl examples) that is not declared elsewhere, and they will transmit user images to an external service (data exfiltration/privacy risk if images are sensitive). The skill does not document data retention, privacy, or what is sent beyond the image URLs.

Install Mechanism

This is an instruction-only skill with no install spec and no code files, so it does not write code to disk or download external binaries. That lowers installation risk.

Credentials

The manifest and SKILL.md expect an API key (Authorization: Bearer {{API_KEY}} / $API_KEY) but requires.env/primary credential fields are empty. This mismatch is problematic: the skill will fail or will silently rely on an implicitly provided key. The skill asks for a high-sensitivity secret (API_KEY) without declaring scope, usage, or least-privilege recommendations.

Persistence & Privilege

always is false and disable-model-invocation is not set; the skill does not request persistent system-wide privileges or modify other skills. There is no install-time persistence specified.

What to consider before installing

This skill will send any submitted image URLs and an API key to an external service (api.ngmob.com). Before installing or using it: (1) ask the author to declare the required env var (API_KEY) in the skill metadata and explain the key's required scope/permissions; (2) verify the service provenance (official homepage, company, support/contact info) and confirm api.ngmob.com is the legitimate endpoint; (3) avoid uploading private/sensitive images; use public or anonymized examples; (4) create a limited-scope API key (least privilege) and monitor its usage; (5) ask the author to document data retention, privacy, and whether images are stored or used to train models; and (6) prefer skills with consistent owner metadata and non-placeholder author fields. The current metadata inconsistencies (missing required credential declaration, no homepage, mismatched owner IDs/placeholder author) are the primary reasons for caution.

Skills.md

Skill Name

virtual-try-on

---

Description

Virtual Try-On — Transform clothing images into professional e-commerce product photos with AI models wearing the garments. Upload up to 4 clothing/garment images and receive a high-quality product photo ready for online retail platforms.

It leverages the Pixify engine to process your inputs through:

• Garment Analysis (image_to_text_gpt5)
• Model Try-On Generation (nano_banana_pro)

---

Service Overview

• 🌐 Product Website / Console:
  https://ai.ngmob.com
  (For product access, workflow management, and obtaining your API Key)

• 🔗 API Base URL:
  https://api.ngmob.com
  (Used strictly for API requests and workflow execution)

---

Use Cases

• Fashion E-Commerce: Generate product photos for online clothing stores
• Design Visualization: See how designs look on models before production
• Catalog Creation: Quickly create professional product catalogs
• Multi-Variant Display: Generate product photo variations
• Retail Preparation: Prepare images for marketplace listings

---

Inputs

Name	Type	Required	Description
Clothing Image 1	string (URL)	✅	Top/Shirt — Clothing item image (shirt, blouse, jacket, etc.)
Clothing Image 2	string (URL)	❌	Bottom/Pants — Clothing item image (pants, skirt, shorts, etc.)
Clothing Image 3	string (URL)	❌	Accessories/Shoes — Accessory or footwear image (shoes, bag, hat, etc.)
Clothing Image 4	string (URL)	❌	Additional Item — Additional clothing or accessory image (optional)

⚠️ Important: Clothing Component Types

• Image 1 (Required): Top/shirt/jacket or any upper body garment
• Image 2 (Optional): Bottom/pants/skirt or any lower body garment
• Image 3 (Optional): Accessories/shoes or footwear
• Image 4 (Optional): Additional clothing items or accessories
• Upload order doesn't matter - workflow automatically identifies and combines components

---

How to Use

When the user requests to execute this workflow, follow these steps:

1. Collect Input Parameters

Gather the required inputs from the user:

• At least 1 clothing/garment image (required)
• Up to 3 additional clothing images (optional)

2. Call the Workflow API

curl -X POST https://api.ngmob.com/api/v1/workflows/2IIk3Z6NKuPZP7moonEI/run \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "inputs": {
      "Clothing Image 1": "https://example.com/shirt.png",
      "Clothing Image 2": "https://example.com/pants.png",
      "Clothing Image 3": "https://example.com/jacket.png",
      "Clothing Image 4": "https://example.com/accessories.png"
    }
  }'

3. Poll Task Status (Recommended: every 3–5 seconds)

Use the returned task_id to query task status:

curl https://api.ngmob.com/api/v1/workflows/executions/{task_id} \
  -H "Authorization: Bearer $API_KEY"

---

Preview

Input Clothing Images

Clothing 1	Clothing 2	Clothing 3	Clothing 4

Generated E-Commerce Product Photos

Product Photo 1	Product Photo 2	Product Photo 3

---

Example (Recommended)

{
  "Clothing Image 1": "https://example.com/shirt.png",
  "Clothing Image 2": "https://example.com/pants.png",
  "Clothing Image 3": "https://example.com/jacket.png",
  "Clothing Image 4": "https://example.com/accessories.png"
}

What happens:

1. The workflow analyzes each clothing image's design and characteristics
2. AI models are dressed with your garments
3. A professional e-commerce product photo is generated
4. You receive an image ready for online retail platforms

---

🤖 Generated with Pixify