kimi-search-mcp
Provides Hermes agents with web search and web page fetching capabilities via the kimi_web_search and kimi_web_fetch tools.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@kimi-search-mcp搜索一下今天 AI 领域的重要新闻"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
kimi-search-mcp
把 Kimi Code 的 web search / fetch 能力包装成标准 MCP Server(stdio),
供 Hermes 等任意 agent 在需要搜索时自动调用。无命令行、无交互,
只有 kimi_web_search 和 kimi_web_fetch 两个工具。
核心调用逻辑参考 Tsukumi233/astrbot_plugin_kimi_web_search,
去掉了 AstrBot 依赖和 /kimi、/kimifetch 命令层。
安装
需要 Python 3.10+,推荐用 uv:
git clone <this-repo> && cd kimi-search-mcp
uv venv .venv && uv pip install -p .venv/bin/python -r requirements.txt
# Windows: uv pip install -p .venv/Scripts/python.exe -r requirements.txt不用 uv 的话,普通 venv + pip 也可以。
Related MCP server: Scout MCP Server
Agent 配置(MCP stdio)
{
"mcpServers": {
"kimi-search": {
"command": "/path/to/kimi-search-mcp/.venv/bin/python",
"args": ["/path/to/kimi-search-mcp/kimi_search_mcp.py"],
"env": {
"KIMI_API_KEY": "sk-..."
}
}
}
}KIMI_API_KEY 必须是 Kimi Code Console 里创建的 API Key(Kimi 会员权益),
不是 platform.moonshot.cn 的开放平台 key。
环境变量
变量 | 默认 | 说明 |
| (必填) | Kimi Code API Key |
|
| 默认返回结果数(1-20) |
|
| 搜索时是否默认抓取页面正文 |
|
| 单条正文截断长度 |
|
| 请求超时(秒) |
| 官方地址 | 端点覆盖,一般不用动 |
|
| 见下文「合规风险」 |
| 自动 | 自定义 User-Agent |
费用
api.kimi.com/coding/v1/search 不是独立计费 API:它走 Kimi 会员的
Kimi Code 额度(订阅制,CLI / VS Code / 第三方工具共享同一额度池,
7 天滚动刷新),没有单独的按次搜索费用。已有会员(Andante ¥49/月起)
搜索就是额度内的既有权益;超额后可开 Extra Usage(按量付费兜底)。
对比:开放平台(platform.moonshot.cn)的内置 web_search 工具按次收费, 本方案不走那条路。
合规风险(重要)
Kimi Code 官方文档明确写道:篡改客户端标识(User-Agent)属于违规,
可能导致会员权益被暂停。本 server 默认以真实身份 kimi-search-mcp/0.1.0
请求。如果服务端校验官方 CLI 身份导致 401/403,可以设
KIMI_CLI_IDENTITY=1 复刻官方 CLI 的 KimiCLI/1.30.0 UA 和 X-Msh-*
请求头(即原插件的做法)——能用,但属于账号持有者自行承担的风险。
测试
已验证(mcp 2.x / Python 3.14):stdio 握手、tools/list、参数校验、
无 key / 错误 key 的报错透传均正常。本仓库附带 smoke_test.py:
# 在仓库根目录、venv 已装依赖后:
.venv/bin/python smoke_test.py .venv/bin/python kimi_search_mcp.py
# 预期输出:TOOLS: ['kimi_web_search', 'kimi_web_fetch'] 及两条中文报错也可以用 MCP Inspector 交互调试:
npx @modelcontextprotocol/inspector .venv/bin/python kimi_search_mcp.py最后给 agent 配好后让它「搜索一下今天的 AI 新闻」,
观察是否自动触发 kimi_web_search。
实现说明
基于 MCP Python SDK v2(
mcp.server.mcpserver.MCPServer),要求mcp>=2.0; v1 的Server+@list_tools装饰器 API 已移除,请勿降级安装。工具内抛
ToolError,错误消息会原文透传给调用方 LLM,便于自我纠错。
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Related MCP Connectors
Scrape, crawl and search the web for AI agents via MCP.
Docs: https://docs.keenable.ai/mcp-server Keenable is a free, remote MCP server that gives agents access to the web index. Search the web with ranked results and date/site filters, then fetch any indexed page as clean markdown. Works out of the box with no account or API key.
Web search, scraping, RAG answers with citations, and translation as MCP tools.
Stealth web browser for agents: search, fetch, click, download and type in persistent MCP sessions.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceWraps the Kimi Coding Search and Fetch APIs into MCP tools for web searching and content retrieval. It enables LLMs to perform targeted searches and crawl web pages using standardized interfaces.1-

Scout MCP Serverofficial
AlicenseNot gradedqualityCmaintenanceEnables web search, scraping, extraction, and crawling through an MCP interface, allowing coding agents to access real-time web data.1MIT- AlicenseAqualityBmaintenanceEnables AI agents to perform multi-engine web search, fetch web pages, and extract clean Markdown content via MCP, with no API keys required.36MIT
- FlicenseNot gradedqualityCmaintenanceBridges MiniMax's native web search server tool to Kimi Code, enabling the agent to perform real-time web searches through an MCP tool.-