weshop-cli

One command. Studio-quality images and videos.

weshop-cli turns WeShop AI into a command-line tool — virtual try-on, model swap, background replace, pose change, canvas expand, AI video generation, video enhancement, and much more. Built for developers and AI agents who want to generate production-ready fashion, product, and creative media without touching a browser.

# Virtual try-on: put a garment onto a model reference
weshop virtualtryon --image ./garment.png --model-image ./model-photo.png --gen-version weshopPro --prompt-mode auto --aspect-ratio 2:3

# Generate a cinematic video from an image using Kling
weshop kling --image ./scene.png --prompt 'Camera slowly pans across a misty forest' --model Kling_3_0 --duration 8s

# Remove background and replace with white
weshop removebg --image ./product.png --mask-type autoSubjectSegment --bg-hex '#ffffff'

Why

• One command = one task. No curl, no JSON, no polling loops.
• Agent-friendly. Structured [key]: value output that any LLM agent can parse. No colors, no spinners, no noise.
• Local files just work. Pass a local path to --image and it auto-uploads. Same file won't upload twice (cached by path + size + mtime).
• Blocking or async. Waits for results by default. Add --no-wait to get the execution ID and poll later with weshop status.

Quick start

Get your API key at open.weshop.ai/authorization/apikey, then:

npm install -g weshop-cli
export WESHOP_API_KEY=<your-key>
weshop --help

For developers

git clone https://github.com/weshopai/weshop-cli.git
cd weshop-cli
npm install
export WESHOP_API_KEY=<your-key>
npx tsx src/index.ts --help

Commands

Run weshop <command> --help to see each command's full parameters, enum values, and examples.

Image Agents

Video Agents

Example: virtual try-on

weshop virtualtryon \
  --image ./garment.png \
  --model-image ./model-photo.png \
  --gen-version weshopPro \
  --prompt-mode auto \
  --aspect-ratio 2:3 \
  --batch 2

Output:

[image]
  imageUrl: https://ai-global-image.weshop.com/...

[submitted]
  executionId: abc123

[result]
  agent: virtualtryon v1.0
  executionId: abc123
  status: Success
  imageCount: 2
  image[0]:
    status: Success
    url: https://ai-global-image.weshop.com/...
  image[1]:
    status: Success
    url: https://ai-global-image.weshop.com/...

Example: AI video generation

# Kling — cinematic video with audio
weshop kling \
  --image ./scene.png \
  --prompt 'Camera slowly pans across a misty forest' \
  --model Kling_3_0 \
  --duration 8s \
  --generate-audio true

# Seedance — ByteDance video model
weshop seedance \
  --image ./photo.png \
  --prompt 'Person walks in slow motion' \
  --model Seedance_15_Pro \
  --duration 8s \
  --aspect-ratio 16:9

# Sora 2 — OpenAI video model
weshop sora-2 \
  --image ./landscape.png \
  --prompt 'Time-lapse of clouds moving' \
  --duration 8s

Example: use preset IDs for best results

For aimodel and aiproduct, using preset scene/model IDs gives the best quality. List them with info:

# See available scenes and models
weshop info aimodel

# Use a preset scene
weshop aimodel \
  --image ./model.png \
  --mask-type autoApparelSegment \
  --generation-mode referToOrigin \
  --location-id 6000372 \
  --batch 2

Async mode

Don't want to wait? Add --no-wait and poll later:

weshop kling --image ./scene.png --prompt 'Ocean waves' --no-wait
# [submitted]
#   executionId: abc123
# [info]
#   message: Use 'weshop status abc123' to check progress

weshop status abc123
# [result]
#   agent: kling v1.0
#   status: Success
#   ...

For AI agents

The output is designed to be easily parsed by automated tools and AI agents:

• Structured [section] + key: value format — no ANSI colors, no progress bars
• Every field is labeled and parseable
• --no-wait + weshop status enables non-blocking workflows
• --help on each command documents every parameter, enum value, and constraint
• Local images are auto-uploaded and cached — no separate upload step needed

License

Apache-2.0