Install
openclaw skills install imgforgeGenerate AI images for free using Z-Image-Turbo. Say "generate an image of..." and get stunning results in seconds.
openclaw skills install imgforgeGenerate images from text using Alibaba's open-source Z-Image-Turbo model. Free to use.
Activate this skill when the user asks to:
python3 SKILL_DIR/imgforge.py "PROMPT" -o OUTPUT_PATH
| Flag | Default | Description |
|---|---|---|
PROMPT (positional) | — | Text description (English or Chinese) |
-o, --out | output.jpg | File path to save the image |
-W, --width | 1024 | Width in pixels (512–2048) |
-H, --height | 1024 | Height in pixels (512–2048) |
--json | off | Output result as JSON |
| Aspect | Flags |
|---|---|
| Square (1:1) | -W 1024 -H 1024 |
| Landscape (16:9) | -W 1280 -H 720 |
| Portrait (9:16) | -W 720 -H 1280 |
| Wide banner | -W 2048 -H 1024 |
Add quality keywords for better output: "4K", "cinematic", "professional photography", "dramatic lighting", "shallow depth of field".
Specify a style: "oil painting", "watercolor", "anime", "pixel art", "minimalist vector".
Requires MODELSCOPE_API_KEY. If the user hasn't set it, guide them:
export MODELSCOPE_API_KEY="ms-..."For Claude Code users, add to ~/.claude/settings.json under "env".
| Symptom | Cause & fix |
|---|---|
| 401 Unauthorized | Token is from modelscope.cn (wrong site) — must use modelscope.ai. Or Alibaba Cloud account not bound. |
| Timeout after 150s | API under load — retry in a minute |
| Content moderation | Rephrase the prompt — Alibaba's filter blocked it |