Web Search MCP server
{
"mcpServers": {
"keenable": {
"type": "http",
"url": "https://api.keenable.ai/mcp"
}
}
}{
"mcpServers": {
"keenable": {
"type": "http",
"url": "https://api.keenable.ai/mcp"
}
}
}{
"mcpServers": {
"keenable": {
"type": "http",
"url": "https://api.keenable.ai/mcp"
}
}
}{
"servers": {
"keenable": {
"type": "http",
"url": "https://api.keenable.ai/mcp"
}
}
}{
"mcpServers": {
"keenable": {
"type": "http",
"url": "https://api.keenable.ai/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsoninstall npx -y @keenable/mcpversion 0.1.0
What it does
A hosted MCP server that exposes live web search and URL-to-markdown fetching via the Keenable web index. It is aimed at developers who want to add web search and page retrieval to AI clients such as Cursor, Cline, VS Code, or Claude Desktop without running their own infrastructure.
- Search the live web with optional filters such as site, date ranges, and search mode to retrieve current information in an AI assistant.
- Fetch any URL and receive its content as clean markdown, suitable for summarisation or further processing by a language model.
- Integrate web search into stdio-only MCP clients like Claude Desktop using the npx wrapper without setting up a local server.
Tools this server exposes
| Tool | Does |
|---|---|
| search_web_pages | Runs a live web search against the Keenable index, with optional filters for site, mode, and publication or acquisition date ranges. |
| fetch_page_content | Fetches a given URL and returns its content as clean markdown, with an optional character limit. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.