Give image recognition ability to models without native vision support (like DeepSeek) — by calling an external visual API.
npx skills add guorui999/visionnode ~/.claude/skills/vision/scripts/vision.js --setupEnter your API Key, API URL, and model name when prompted.
Or edit config directly: ~/.claude/skills/vision/config.json
{
"base_url": "https://dashscope.aliyuncs.com/compatible-mode/v1",
"api_key": "your-api-key",
"model": "qwen3.5-omni-plus"
}After setup, send an image or say "look at this image" — Claude will auto-trigger the skill.
# Local image
node scripts/vision.js photo.jpg "Describe this image"
# Remote image
node scripts/vision.js --url https://example.com/img.png "What is this?"
# Multiple images
node scripts/vision.js img1.jpg img2.jpg "Compare these two"
# With cache (skip API call if same image + prompt analyzed before)
node scripts/vision.js --cache photo.jpg "Describe this image"jpg, jpeg, png, gif, webp, bmp
| Service | Model | Notes |
|---|---|---|
| Alibaba Cloud Bailian | qwen3.5-omni-plus |
1M free tokens for new users |
| Alibaba Cloud Bailian | qwen-vl-max |
Same as above |
| OpenAI | gpt-4o-mini |
Requires overseas payment |
| Others | Any OpenAI-compatible | Change BASE_URL + model name |
- Read image → base64 encode
- Call visual API (OpenAI-compatible format)
- Return text description
- SKILL.md — Full documentation (CN/EN)
- install.sh — One-click install script
让没有原生识图能力的模型(如 DeepSeek)也能"看图"——通过调用外部视觉 API 获取图片的文字描述。
npx skills add guorui999/visionnode ~/.claude/skills/vision/scripts/vision.js --setup按提示输入 API Key、API 地址、模型名称。
或直接编辑配置文件:~/.claude/skills/vision/config.json
{
"base_url": "https://dashscope.aliyuncs.com/compatible-mode/v1",
"api_key": "你的API Key",
"model": "qwen3.5-omni-plus"
}配置完成后,直接发送图片或说"看这张图"即可自动识别。
# 本地图片
node scripts/vision.js photo.jpg "描述这张图片"
# 网络图片
node scripts/vision.js --url https://example.com/img.png "这是什么?"
# 多张图片
node scripts/vision.js img1.jpg img2.jpg "比较这两张图"
# 缓存模式(同一图片+prompt 避免重复调用 API)
node scripts/vision.js --cache photo.jpg "描述这张图片"jpg, jpeg, png, gif, webp, bmp
| 服务 | 模型 | 备注 |
|---|---|---|
| 阿里云百炼(推荐) | qwen3.5-omni-plus |
新用户 100 万 token 免费 |
| 阿里云百炼 | qwen-vl-max |
同上 |
| OpenAI | gpt-4o-mini |
需海外支付 |
| 其他 | 任何 OpenAI 兼容格式 | 改 BASE_URL 和模型名即可 |
- 读取图片文件 → 转换为 base64
- 调用视觉 API(OpenAI 兼容格式)
- 返回文字描述
- SKILL.md — 完整使用说明(中英双语)
- install.sh — 一键安装脚本