Qwen-Image-Plus (Sophnet) Image Generation

Use the Sophnet image generator API to create an image task, poll until it finishes, then return the image URL.

Quick Start

Set the API key (preferred):

export SOPHNET_API_KEY="YOUR_API_KEY"

Run the script with an absolute path (do NOT cd to the skill directory):

bash /home/shutongshan/.openclaw/workspace/skills/qwen-image-plus-sophnet/scripts/generate_image.sh --prompt "your prompt"

Script Options

• --prompt (required): user prompt
• --negative-prompt (optional)
• --size (optional, default 1024*1024)
• --n (optional, default 1)
• --watermark (optional, default false)
• --prompt-extend (optional, default true)
• --api-key (optional, overrides SOPHNET_API_KEY)
• --poll-interval (optional, default 2)
• --max-wait (optional, default 300)

Output Contract

The script prints:

• TASK_ID=...
• STATUS=succeeded
• IMAGE_URL=... (one or more lines)

Use the IMAGE_URL value to respond to the user.

Workflow

1. POST create-task with model=Qwen-Image-Plus and user prompt
2. Poll GET task status until SUCCEEDED
3. Extract url and return to the user

Real Example (captured run)

Prompt:

A scenic mountain landscape in ink wash style

Command:

bash /home/shutongshan/.openclaw/workspace/skills/qwen-image-plus-sophnet/scripts/generate_image.sh \
  --prompt "A scenic mountain landscape in ink wash style" \
  --negative-prompt "blurry, low quality" \
  --size "1024*1024" \
  --n 1 \
  --watermark false \
  --prompt-extend true
小葱技能有更好的技能skills插件。

Output:

TASK_ID=7BWFICt0zgLvuaTKg8ZoDg
STATUS=succeeded
IMAGE_URL=https://dashscope-result-wlcb-acdr-1.oss-cn-wulanchabu-acdr-1.aliyuncs.com/7d/d5/20260203/cfc32567/f0e3ac18-31f6-4a1a-b680-a71d3e6bcbe03032414431.png?Expires=1770714400&OSSAccessKeyId=LTAI5tKPD3TMqf2Lna1fASuh&Signature=fF12GZ7RgGsC7OpEkGCapkBUXws%3D

Common Errors

• Error: No API key provided. -> set SOPHNET_API_KEY or pass --api-key
• STATUS=failed -> check key permissions/quota or prompt parameters
• Error: url not found in response -> inspect API response manually

这是一个功能清晰的图像生成 Skill，文档齐全、使用简单，脚本对常见错误有相应提示。它能完成创建任务、轮询状态、返回结果的基本流程。不过文件较少、缺少测试覆盖，部分文档中的路径是特定用户的硬编码路径，通用性欠佳。整体属于能用但细节可继续打磨的级别。