Grok Search MCP

Grok Search MCP

A local stdio MCP server that exposes xAI-compatible server-side search through a Responses API relay, providing web and X search tools with citation and cost reporting.

Category
访问服务器

README

Grok Search MCP

A local stdio MCP server that exposes xAI-compatible server-side search through a Responses API relay.

Tools

  • grok_web_search: public web search with optional allowed or excluded domains.
  • grok_x_search: X post/profile/thread search. This is deliberately separate and marked high-cost because one request can expand into many metered tool calls.

Both tools report citations, actual upstream tool-call counts, token usage, the effective model, and the relay's raw cost_in_usd_ticks value when present.

Configuration

Required at call time:

BOTCF_API_KEY

On Windows, run npm run setup for the first-install configuration window. Enter Base URLs, models, and API keys separated by semicolons; the positions are paired, and one model value may be reused for every provider. Public provider settings are stored in config.json, while the complete provider array including keys is stored as a current-user DPAPI ciphertext under %APPDATA%\grok-search-mcp\api-key.dpapi.

Requests rotate across configured providers. Recoverable network/timeout, authentication, rate-limit, route/model compatibility, and 5xx failures can switch to the next distinct provider. Each provider is attempted at most once and each MCP call has a hard maximum of three provider attempts. A failover can still duplicate upstream work if a provider accepted a request before its response was lost, so account-side spend limits remain necessary.

For CI or temporary sessions, GROK_SEARCH_PROVIDERS may contain a JSON provider array. The legacy BOTCF_API_KEY, BOTCF_BASE_URL, and GROK_SEARCH_MODEL variables remain supported.

Optional:

BOTCF_BASE_URL=https://botcf.com
GROK_SEARCH_MODEL=grok-4.5

The server starts and lists tools without a key, then fails closed with a clear error if a tool is called before BOTCF_API_KEY is configured. Secrets are never accepted as tool arguments or printed.

Each installation must use its own provider credentials. Do not share .env files, DPAPI ciphertext, setup screenshots, or API keys with other users.

Development

npm install
npm run check

Codex configuration

After building, register the compiled entry point and forward only the required environment variable:

[mcp_servers.grok_search]
command = "node"
args = ["D:/path/to/grok-search-mcp/dist/index.js"]
env_vars = ["GROK_SEARCH_PROVIDERS", "BOTCF_API_KEYS", "BOTCF_API_KEY", "GROK_SEARCH_BASE_URLS", "BOTCF_BASE_URL", "GROK_SEARCH_MODELS", "GROK_SEARCH_MODEL"]
startup_timeout_sec = 20
tool_timeout_sec = 180

No request is retried automatically. HTTP redirects are rejected so the API key cannot be forwarded to another origin.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选