Novada MCP
Live on Vercel · 8 core tools Vercel 部署 · 8 个核心工具

One URL.
8 core data tools.
Zero install.
一个 URL。
8 个核心数据工具。
零安装。

Search, scrape, extract, crawl, map, research, verify — through a single MCP endpoint. Drop into any AI client in 30 seconds. 搜索、抓取、提取、爬取、站点地图、研究、验证 — 全部通过一个 MCP 端点。30 秒接入任意 AI 客户端。

1,000 free calls/month · No credit card · Claude, Cursor, Codex, Cline, Windsurf 每月 1,000 次免费调用 · 无需信用卡 · Claude、Cursor、Codex、Cline、Windsurf

tools/list response
LIVE
// 8 core tools available
{ "tools": [
"novada_search" — web search via 5 engines
"novada_extract" — read content from URL
"novada_crawl" — multi-page site crawl
"novada_scrape" — structured platform data
"novada_research" — deep multi-source analysis
"novada_verify" — fact-check claims
"novada_map" — discover site URLs
"novada_monitor" — track page changes
] }
Streamable HTTP · JSON-RPC 2.0
8
core tools核心工具
195+
countries国家覆盖
100M+
proxy IPs代理 IP

Install in 30 seconds 30 秒接入

Pick your client. Copy the snippet. Replace YOUR_API_KEY with your real key. 选择客户端,复制片段,把 YOUR_API_KEY 替换为你的真实密钥。

Option A -- Custom Connector UI (Free, Pro, Max). Go to Customize > Connectors > + > Add custom connector and paste the URL. 方式 A -- 自定义连接器界面(Free、Pro、Max 均可)。进入 Customize > Connectors > + > Add custom connector,粘贴下方 URL。

https://mcp.novada.com/YOUR_API_KEY/mcp

Option B -- stdio bridge. Edit claude_desktop_config.json: 方式 B -- stdio 桥接。编辑 claude_desktop_config.json:

{
  "mcpServers": {
    "novada": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.novada.com/mcp?token=YOUR_API_KEY"
      ]
    }
  }
}

Config: ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows). 配置文件:~/Library/Application Support/Claude/claude_desktop_config.json (macOS) 或 %APPDATA%\Claude\claude_desktop_config.json (Windows)。

Recommended -- pass the key as a header, scope to user so every project sees it: 推荐方式 -- 使用请求头传递密钥,设为用户级作用域:

$ claude mcp add --transport http novada \
    https://mcp.novada.com/mcp \
    --header "Authorization: Bearer YOUR_API_KEY" \
    --scope user

Verify with claude mcp list. Without --scope user the server is registered only in the current project. claude mcp list 验证。不加 --scope user 时仅当前项目可用。

Add to ~/.cursor/mcp.json: 添加到 ~/.cursor/mcp.json:

{
  "mcpServers": {
    "novada": {
      "url": "https://mcp.novada.com/mcp?token=YOUR_API_KEY"
    }
  }
}

After saving, open Cursor Settings > Features > Model Context Protocol and confirm Novada toggles green. 保存后打开 Cursor Settings > Features > Model Context Protocol,确认 Novada 显示为绿色。

Add to ~/.codex/config.toml: 添加到 ~/.codex/config.toml:

[mcp_servers.novada]
url = "https://mcp.novada.com/mcp?token=YOUR_API_KEY"

To keep the key out of the file, use bearer_token_env_var = "NOVADA_API_KEY" against https://mcp.novada.com/mcp. 若不想将密钥写入文件,可改用 bearer_token_env_var = "NOVADA_API_KEY" 搭配 https://mcp.novada.com/mcp

Edit cline_mcp_settings.json (Cline > MCP Servers > Edit): 编辑 cline_mcp_settings.json(Cline > MCP Servers > 编辑):

{
  "mcpServers": {
    "novada": {
      "type": "streamableHttp",
      "url": "https://mcp.novada.com/mcp?token=YOUR_API_KEY"
    }
  }
}

"type": "streamableHttp" must be explicit -- without it, Cline defaults to SSE. "type": "streamableHttp" 必须显式写出,否则 Cline 默认走 SSE。

Edit ~/.codeium/windsurf/mcp_config.json: 编辑 ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "novada": {
      "serverUrl": "https://mcp.novada.com/mcp?token=YOUR_API_KEY"
    }
  }
}

Windsurf uses serverUrl (url also works). After saving, press the refresh button in the MCP panel. Windsurf 使用 serverUrl(也兼容 url)。保存后在 MCP 面板按刷新。


Agent frameworks Agent 框架集成

Point any MCP-aware framework at the same endpoint. 让任意支持 MCP 的框架指向同一端点。

Python -- pip install langchain-mcp-adapters Python -- pip install langchain-mcp-adapters

from langchain_mcp_adapters.client import MultiServerMCPClient

client = MultiServerMCPClient({
    "novada": {
        "transport": "http",
        "url": "https://mcp.novada.com/mcp",
        "headers": {"Authorization": "Bearer YOUR_NOVADA_API_KEY"},
    }
})
tools = await client.get_tools()

Python -- pip install llama-index-tools-mcp Python -- pip install llama-index-tools-mcp

from llama_index.tools.mcp import BasicMCPClient, McpToolSpec

client = BasicMCPClient(
    "https://mcp.novada.com/mcp",
    headers={"Authorization": "Bearer YOUR_NOVADA_API_KEY"},
)
tools = await McpToolSpec(client=client).to_tool_list_async()

Python -- pip install openai-agents Python -- pip install openai-agents

from agents.mcp import MCPServerStreamableHttp

server = MCPServerStreamableHttp(
    name="Novada MCP",
    params={
        "url": "https://mcp.novada.com/mcp",
        "headers": {"Authorization": "Bearer YOUR_NOVADA_API_KEY"},
    },
    cache_tools_list=True,
)

Python -- pip install crewai-tools Python -- pip install crewai-tools

from crewai_tools import MCPServerAdapter

params = {
    "url": "https://mcp.novada.com/mcp",
    "transport": "streamable-http",
    "headers": {"Authorization": "Bearer YOUR_NOVADA_API_KEY"},
}
with MCPServerAdapter(params) as tools:
    agent = Agent(..., tools=tools)

MCP Client Tool node. Bridge Streamable-HTTP via mcp-remote: MCP Client Tool 节点。用 mcp-remote 桥接:

$ npx mcp-remote https://mcp.novada.com/mcp \
    --header "Authorization: Bearer YOUR_NOVADA_API_KEY"

Then set the node's SSE Endpoint to the local bridge URL (e.g. http://localhost:3000/sse). 然后在节点的 SSE Endpoint 填本地桥接地址(如 http://localhost:3000/sse)。


Why hosted MCP? 为什么用托管 MCP?

Zero install零安装

Just paste a URL. No npm, no env vars, no rebuilds. Works on any machine in 30 seconds. 只需粘贴一个 URL,无需 npm、无需环境变量。任何机器 30 秒接入。

Always latest始终最新

Tools update server-side. New features and fixes land without a redeploy on your side. 工具在服务端更新,新功能和修复无需你重新部署即可生效。

Edge-deployed边缘部署

Runs on Vercel with global PoPs. Latency tracks the user, not the data center. 运行于 Vercel,具备全球节点。延迟随用户位置就近响应。


Copied已复制