妙达图片理解

Use when user needs to understand, analyze, or describe image content. Do NOT use Read tool to read images — use this skill instead. Supports local file path...

Security Scan

The skill's instructions match an image-understanding purpose but they require an external CLI ('miaoda-studio-cli') that is not declared or provided, which is an incoherence and a potential privacy/network risk.

Purpose & Capability

The skill states its purpose is image understanding and the SKILL.md shows only that behavior, but it instructs use of the external binary 'miaoda-studio-cli' while the skill metadata declares no required binaries or install steps. That mismatch (instructions depending on a CLI that isn't declared or installed) is incoherent and could hide implicit dependencies.

Instruction Scope

Instructions limit actions to processing a local image path and optional prompts (describe, extract text, etc.) and do not ask the agent to read unrelated files or env vars. However, the CLI call could perform network I/O (uploading images) or access other system resources — the SKILL.md does not state what miaoda-studio-cli does with the image data.

Install Mechanism

There is no install spec and no code files — this is instruction-only. That is low-risk in general, but here it's problematic because the instructions rely on an external tool that the package does not declare or provide; the skill should either declare the required binary or include an install step.

Credentials

The skill requests no environment variables, credentials, or config paths, which is appropriate for a simple image-analysis instruction-only skill.

Persistence & Privilege

The skill does not request persistent/always-on presence and has default invocation settings; it does not attempt to modify other skills or system-wide settings.

What to consider before installing

This skill's instructions call an external CLI ('miaoda-studio-cli') but the skill package doesn't declare or install that binary and provides no homepage or source. Before installing or using it, verify where 'miaoda-studio-cli' comes from and whether you already trust it: check the binary's origin, source code or vendor, and its network behavior (does it upload images?). If you care about privacy, test with non-sensitive images first or run the CLI offline in a controlled environment. Prefer skills that declare required binaries or include an install spec and provide a homepage/source for auditing.

Image Understanding

通过 miaoda-studio-cli image-understanding 使用 AI 理解、分析和描述图片内容。

Quick Reference

参数	说明	必需	默认值
--image, -i	本地图片文件路径	是	-
--prompt, -p	关于图片的问题或指令	否	请描述这张图片的内容
--output, -o	输出格式: text/json	否	text

使用示例

# 基础用法：描述图片内容
miaoda-studio-cli image-understanding --image ./photo.png

# 针对图片提问
miaoda-studio-cli image-understanding --image ./photo.png --prompt "这张图片中有什么"

# 分析本地图片并输出 JSON
miaoda-studio-cli image-understanding --image ./screenshot.png --prompt "提取图片中的文字" --output json

# 简写参数
miaoda-studio-cli image-understanding -i ./photo.jpg -p "分析图片的色彩风格"

使用场景决策

需要理解图片内容
├─ 描述图片整体内容 → 不指定 --prompt，使用默认描述
├─ 针对图片提问 → 用 --prompt 指定具体问题
├─ 提取图片中的文字 → --prompt "提取图片中的文字"
├─ 分析图片风格/构图 → --prompt 描述分析需求
└─ 需要程序处理结果 → 加 --output json

Common Mistakes

错误	正确做法
本地文件路径包含空格未加引号	路径有空格时用引号包裹：-i "my image.png"
用此命令处理文档（PDF/Word）	文档解析用 miaoda-studio-cli doc-parse（见 miaoda-doc-parse skill）
prompt 过于模糊（如"分析"）	明确指定需求，如"提取文字"、"描述场景"、"识别物体"