Qwen Vision

Analyze images and videos using Qwen Vision API (Alibaba Cloud DashScope). Supports image understanding, OCR, visual reasoning.

Security Scan

The skill's files and instructions are consistent with its stated purpose (sending images to Alibaba's DashScope/Qwen Vision API) and request only the expected API key and python3; minor documentation vs. code mismatches are present but not suspicious.

Purpose & Capability

Name/description match the actual behavior: the included script base64-encodes an image and posts it to DashScope's Qwen Vision endpoint. The only required binary is python3, which is reasonable.

Instruction Scope

SKILL.md instructs running the shipped script and documents possible locations to obtain the API key (including ~/.openclaw/openclaw.json or DASHSCOPE_API_KEY). The runtime script itself, however, requires an --api-key argument and does not read those config paths or environment variables. The instructions do not direct reading unrelated files or exfiltrating system data.

Install Mechanism

No install spec or remote downloads are present; this is an instruction-only skill with a small included script. Nothing is fetched from external/untrusted URLs during install.

Credentials

The skill legitimately requires a DashScope API key. Registry metadata lists no required env vars, but SKILL.md suggests DASHSCOPE_API_KEY or ~/.openclaw config as sources; the script actually requires an explicit --api-key. This is an implementation/documentation mismatch but the credential requested is proportional to the skill's function.

Persistence & Privilege

Skill is not always-enabled, does not request persistent or elevated system privileges, and does not modify other skills or system-wide configuration.

Assessment

This skill sends your image (base64-encoded) and prompt to Alibaba Cloud DashScope (Qwen Vision) — you'll need a valid API key. Confirm you trust the external service before sending sensitive images. Note a small mismatch: documentation suggests the key can be read from ~/.openclaw/openclaw.json or DASHSCOPE_API_KEY, but the bundled script requires --api-key (it doesn't auto-read those locations). If you want the convenience of env/config lookup, verify or modify the script yourself. Otherwise, review where you store the API key and avoid passing it in shells or logs if you need secrecy. Lastly, since the skill transmits image data externally, ensure this aligns with your privacy/compliance requirements.

Analyze images and videos using Alibaba Cloud's Qwen Vision API (通义千问视觉模型).

Usage

Analyze an image:

uv run {baseDir}/scripts/analyze_image.py --image "/path/to/image.jpg" --prompt "请描述这张图片" --api-key sk-xxx

With custom model:

uv run {baseDir}/scripts/analyze_image.py --image "/path/to/image.jpg" --model qwen-vl-max-latest --api-key sk-xxx

API Key

Get your API key from:

• models.providers.bailian.apiKey in ~/.openclaw/openclaw.json
• Or skills."qwen-image".apiKey in ~/.openclaw/openclaw.json
• Or DASHSCOPE_API_KEY environment variable
• Or https://dashscope.console.aliyun.com/

Models

Model	Description
qwen-vl-max-latest	Latest max model (default)
qwen-vl-plus-latest	Faster, cost-effective

Prompt Examples

Task	Prompt
Describe	"请详细描述这张图片的内容"
OCR	"提取图片中的所有文字"
Count	"数一下图中有多少个物体"
Analyze	"分析这张图表的数据趋势"
Identify	"这是什么地方/物品？"

Notes

• Supports JPG, PNG, GIF, WebP, BMP formats
• Images are encoded as base64 and sent via API
• Response time varies by image size and complexity