/alibabacloud-bailian-image-creator

name: alibabacloud-bailian-image-creator
description: AI image creation skill supporting text-to-image, image editing, image understanding, and more. Uses qwen-image-2.0-pro, wan2.7-image, qwen3.5-plus and other models. Use this skill when users need to generate images, edit images, analyze image content, or perform image-related tasks. Note: on first run, it will auto-manage DashScope API Keys (create/recycle) and may auto-install the Alibaba Cloud CLI ModelStudio plugin.

AI Image Creator

Professional-grade AI image creation skill built on Alibaba Cloud DashScope API.

> **Warning: Before executing any image task, you must read and follow the "Mandatory Rules" section. Violations will cause task failure.**

Mandatory Rules

First Principle: Must Use Existing Scripts

**All image generation, editing, and understanding tasks must and can only be completed by running existing scripts in the `scripts/` directory.**

• **Do NOT write your own API call code** -- Do not create new Python scripts to call DashScope API. You must use the scripts specified in the "Task-to-Script Mapping" table below
• **Do NOT use PIL/Pillow or other local libraries** for image generation or editing (only allowed for auxiliary operations such as size validation)
• **Do NOT use third-party APIs** -- Including but not limited to Pollinations.ai, Stability.ai, DALL-E, Midjourney
• **Do NOT create mock/simulated scripts** -- Do not bypass real API calls in any way
• **Do NOT use deprecated models** -- `wanx-v1`, `wanx-v2`, `wan2.6-image` and other legacy models are discontinued
• **Do NOT mix APIs** -- Qwen models (qwen-*) can only use `MultiModalConversation.call()`, Wanx models (wan2.7-*) can only use `ImageGeneration.call()`
• **Do NOT use curl to call DashScope REST API** -- Must call through provided scripts, do not write shell scripts as substitutes
• **Do NOT generate placeholder/blank images** -- Do not use hardcoded byte streams, blank canvases, or PIL-scaled images to impersonate API-generated images
• **Do NOT claim success when API calls fail** -- Must truthfully report failure reasons, do not generate false success reports

Task-to-Script Mapping (The Only Execution Method)

Based on user request keywords, look up the table to select the script. **No other methods are allowed**:

| User Request Keywords | Script | API | Default Model | |----------------------|--------|-----|---------------| | Wanx, wan2.7, reference image + generation, style fusion, multi-image fusion, graffiti-on-car, virtual try-on | `wanx_generate.py` | `ImageGeneration.call()` | `wan2.7-image` | | 4K HD text-to-image | `wanx_generate.py` | `ImageGeneration.call()` | `wan2.7-image-pro` | | A set of coherent images, image series generation | `wanx_generate.py` | `ImageGeneration.call()` | `wan2.7-image-pro` | | Text-to-image, generate image, draw a picture (no reference image) | `text_to_image.py` | `MultiModalConversation.call()` | `qwen-image-2.0-pro` | | Edit image (URL input) | `image_edit.py` | `MultiModalConversation.call()` | `qwen-image-edit-max` | | Edit image (local file) | `image_edit_base64.py` | `MultiModalConversation.call()` | `qwen-image-edit-max` | | Analyze/understand image content, describe image | `image_understanding.py` | `chat.completions.create()` | `qwen3.5-plus` |

**Disambiguation Priority Rules** (when multiple keywords match simultaneously): 1. User mentions "Wanx" or "wan2.7" -> Use `wanx_generate.py` directly, regardless of whether "generate image" or similar words are present 2. User provides **reference image URLs** and requests generating a new image -> Use `wanx_generate.py` (Wanx supports multi-image reference input) 3. Text description only, no reference image -> Use `text_to_image.py`

**Allowed Models**:

| Script | Allowed Models | |--------|---------------| | `text_to_image.py` | `qwen-image-2.0-pro`, `qwen-image-2.0` | | `wanx_generate.py` | `wan2.7-image-pro`, `wan2.7-image` | | `image_edit.py` / `image_edit_base64.py` | `qwen-image-edit-max`, `qwen-image-2.0-pro`, `qwen-image-edit-plus`, `qwen-image-edit` | | `image_understanding.py` | `qwen3.5-plus`, `qwen-vl-max`, `qwen-vl-plus` |

API Key Security Management

**Scripts automatically handle key retrieval via `api_key.py`. The Agent does not need to and should not manually extract, set, or pass API Key values.**

1. **Key retrieval is automated**: Scripts internally call `api_key.py` to automatically obtain keys from config files/environment variables, or auto-create via CLI. The Agent only needs to run the script command 2. **Never hardcode any form of key**: Including `api_key = "sk-..."`, `export DASHSCOPE_API_KEY="sk-..."`, and assigning keys in shell scripts 3. **Never extract keys from CLI output**: The key value returned by `aliyun modelstudio create-api-key` is automatically saved by `api_key.py`. The Agent must not write this value into any script, variable, or file 4. **Never expose keys in any output**: Including generated scripts, shell commands, log files (summary.md, task_summary.md, execution_log.md, etc.), and terminal output containing strings starting with `sk-` 5. **Never read or print keys from config files**: Do not use `cat ~/.aliyun/config.json`, `jq`, `python -c`, or other commands to read and output API Key values from config files 6. **Never directly call CLI key commands**: Do not directly run `aliyun modelstudio create-api-key` or `aliyun modelstudio list-api-keys`. These commands are only called internally by `api_key.py` 7. **Mandatory self-check before task completion**: Run `grep -rn "sk-" <output_directory>/` to check all output files; if any strings starting with `sk-` are found (excluding "sk-xxx" placeholders), delete the affected files and regenerate

API Call Result Validation

1. **Check status code**: API response `status_code` must be `200`, otherwise abort the task 2. **Check output content**: Response must contain valid image URLs (starting with `http`). No valid URL me