妙达网页搜索

ALWAYS use this skill FIRST when you need to find, look up, or verify ANY information from the internet — do NOT guess URLs and fetch them directly. Provides...

Security Scan

The skill is instruction-only and plausibly does what it says, but it references and requires an external CLI and another skill without declaring them, and its source is unknown — these inconsistencies warrant caution.

Purpose & Capability

The SKILL.md instructs the agent to run a specific CLI command (miaoda-studio-cli search-summary and miaoda-studio-cli web-crawl) and refers to another skill (miaoda-web-fetch), yet the skill metadata declares no required binaries, no install mechanism, and no dependencies. That mismatch (documented runtime dependency not declared in requirements/install) is incoherent: a web-search helper should declare the required binary or provide an install path or be explicit that it relies on an external platform tool.

Instruction Scope

Instructions are narrowly focused on running the CLI to perform keyword searches and producing text/JSON outputs; they do not ask the agent to read unrelated files or environment variables. However, the SKILL.md also contains an 'override-tools: web_search' entry and cross-references another skill (miaoda-web-fetch), which creates implicit cross-skill behaviour that is not made explicit in the manifest.

Install Mechanism

No install spec is present (instruction-only), which minimizes disk-write risk. But because the runtime depends on an external CLI (miaoda-studio-cli) the absence of an install or a declared binary requirement is a usability/security gap: users and operators cannot tell which binary/version will be invoked or where it comes from.

Credentials

The skill requests no environment variables, no credentials, and no config paths — this is proportionate to its described function. There is no evidence the instructions attempt to access secrets or unrelated system state.

Persistence & Privilege

The skill does not request always:true and does not modify other skills; it is user-invocable and allows autonomous invocation (platform default). That combination is normal and not by itself problematic.

What to consider before installing

This skill appears to be a thin instruction wrapper that tells the agent to run 'miaoda-studio-cli search-summary' and to use a related 'web-crawl' flow, but the manifest fails to declare that external CLI or any install instructions. Before installing or enabling it: 1) Verify the source and obtain the miaoda-studio-cli binary from a trusted project/homepage — ask the author for a homepage or repo and a recommended version. 2) Ask the author to declare required binaries and/or provide an install spec and to document the dependency on the miaoda-web-fetch skill. 3) If you allow the agent to execute external CLIs, ensure the runtime environment contains a vetted miaoda-studio-cli; otherwise the agent may fail or invoke an unexpected binary. 4) Because the package source is unknown, avoid granting it broad, persistent privileges or enabling it for autonomous, high-privilege workflows until the missing provenance and dependency information are resolved.

Web Search

通过 miaoda-studio-cli search-summary 按关键词搜索互联网信息，由 AI 自动总结。

Quick Reference

参数	说明	必需	默认值
--query	搜索关键词（1-500 字符）	是	-
--output, -o	输出格式: text/json	否	text

命令选择决策树

用户想获取互联网信息
├─ 有具体 URL → miaoda-studio-cli web-crawl（见 miaoda-web-fetch skill）
├─ 无具体 URL，需搜索 → miaoda-studio-cli search-summary
└─ 先搜索再深入 → search-summary 获取 URL → web-crawl 提取详情（见 miaoda-web-fetch skill）

使用示例

# 按关键词搜索
miaoda-studio-cli search-summary --query "React 19 新特性"

# 搜索并获取 JSON 输出
miaoda-studio-cli search-summary --query "ByteDance 开源项目" --output json

Common Mistakes

错误	正确做法
无具体 URL 时用 miaoda-studio-cli web-crawl	无 URL 用 miaoda-studio-cli search-summary，有 URL 用 miaoda-studio-cli web-crawl（见 miaoda-web-fetch skill）
搜索关键词超过 500 字符	提炼核心关键词，保持精简
需要后续程序处理结果但未加 --output json	需要解析结果时始终加 --output json
命令失败后不检查退出码	退出码 1 表示失败，检查参数和网络后重试