glm-web-search

使用 GLM 联网搜索 MCP 进行网络搜索。触发条件：(1) 用户要求进行网络搜索、在线搜索、查找信息 (2) 需要查询最新资讯、新闻、资料 (3) 使用 GLM 的 web_search 功能

Security Scan

The skill's instructions, requested data, and behavior are coherent with a GLM-based web search helper, but it asks you to run an npx package and write a local API key file and has small endpoint/auth inconsistencies you should verify before use.

Purpose & Capability

The skill name/description (GLM web search) align with its actions: it asks for a GLM/智谱 API key, configures a local MCP entry, and calls a web_search MCP tool. Requiring a GLM API key and using an MCP client (mcporter) is expected for this purpose.

Instruction Scope

Instructions stay within the stated purpose (install/use mcporter, store/read a GLM API key in ~/.openclaw/config/glm.json, add and call an MCP server). Two minor inconsistencies: one mcporter URL example uses an SSE endpoint with Authorization as a query parameter, while the MCP configuration section lists a different mcp URL and says Bearer token in headers — you should confirm which form the service expects. The skill instructs writing the API key to a local file, which is expected but worth knowing.

Install Mechanism

No formal install spec in the registry, but the SKILL.md tells users/agents to run 'npx -y mcporter' which fetches and executes a package from the npm registry. This is functionally reasonable for using mcporter but carries the usual moderate risk of executing remote npm code at runtime — verify the mcporter package and publisher before running.

Credentials

The skill requests no environment variables and only asks to read/write a single local config file (~/.openclaw/config/glm.json) to store the GLM API key. That is proportional to the stated purpose. It does not ask for unrelated credentials or broad access.

Persistence & Privilege

The skill does not request always:true or other elevated platform privileges. Its only persistent action is writing the GLM API key to a local config file under ~/.openclaw, which is within scope for a tool that needs a local API key.

Assessment

This skill appears to do what it says: configure and call GLM's web_search via mcporter. Before installing/using it: 1) Confirm you trust the mcporter npm package and its publisher because the instructions use 'npx' which will fetch and run code from the registry. 2) Verify the exact MCP endpoint and auth method (the SKILL.md shows both an SSE URL with Authorization in the query and a separate mcp URL that expects a Bearer header) and adjust the config so your key is sent securely (prefer Authorization header over embedding secrets in URLs). 3) Be aware the skill will create/modify ~/.openclaw/config/glm.json to store your GLM API key — if you prefer, create that file yourself with the key rather than pasting it into interactive prompts. 4) Only provide your GLM/智谱 API key if you trust the BigModel service and this skill's source; obtain the key from the official site. If you want higher assurance, ask the skill author for the mcporter package name/version they expect and inspect that package before running it.

使用 GLM 联网搜索 MCP 服务器进行网络搜索。

执行流程（首次需要安装，后续直接步骤6调用）

步骤 1: 检查并安装依赖

1.1 检查 mcporter 是否可用

npx -y mcporter --version

如果命令返回成功，说明 mcporter 可用，跳到步骤 2。

mcporter 可以直接通过 npx 使用，无需安装。

步骤 2: 检查 API Key 配置

cat ~/.openclaw/config/glm.json 2>/dev/null | python3 -c "import json,sys; d=json.load(sys.stdin); print(d.get('api_key', ''))"

如果返回非空的 API Key，跳到步骤 4。

步骤 3: 配置 API Key（如果未配置）

3.2 如果没有找到 Key，向用户索要

如果用户没有智谱 API Key，可以访问 https://www.bigmodel.cn/glm-coding?ic=OOKF4KGGTW 购买。

询问用户提供智谱 API Key。

3.3 保存 API Key

mkdir -p ~/.openclaw/config
cat > ~/.openclaw/config/glm.json << EOF
{
  "api_key": "API密钥"
}
EOF

步骤 4: 添加 MCP 服务器

使用 mcporter 添加 GLM 联网搜索 MCP 服务器：

mcporter config add glm-search \
  --type sse \
  --url "https://open.bigmodel.cn/api/mcp/web_search_prime/sse?Authorization=your-key"

注意：将 your-key 替换为实际的智谱 API Key。

步骤 5: 测试连接

mcporter list

确认 glm-search 服务器已成功添加。

步骤 6: 使用 mcporter 调用 MCP 进行网络搜索

6.1 使用 mcporter 调用 MCP 工具

使用 mcporter 调用 MCP 服务：

mcporter call glm-search.webSearchPrime search_query="<搜索查询>"

示例：

# 搜索今日新闻
mcporter call glm-search.webSearchPrime search_query="今天的热点新闻"

# 搜索技术信息
mcporter call glm-search.webSearchPrime search_query="Python 异步编程最佳实践"

6.2 API 参数说明

参数	说明	类型
search_query	搜索查询字符串	string (必填)

支持的工具

重要提示：如果出现问题以官方说明为准 官方版说明 ： https://docs.bigmodel.cn/cn/coding-plan/mcp/search-mcp-server

GLM 联网搜索 MCP 服务器提供以下工具：

• webSearchPrime - 搜索网络信息，返回结果包括网页标题、网页URL、网页摘要、网站名称、网站图标等

MCP 配置

MCP 服务器名称：glm-search

MCP 服务器类型：HTTP

MCP 服务器 URL：https://open.bigmodel.cn/api/mcp/web_search_prime/mcp

认证方式：Bearer Token（Authorization header）