SuperAPI AI Directory

keenable-mcp

🎖️ 📇 ☁️ - Live web search and clean Markdown page fetching over the Keenable web index.

Open website

About this tool

Keenable MCP is a server that provides real-time web search and page content fetching capabilities for MCP (Model Context Protocol) clients, running on top of the Keenable web index.

Purpose This service is mainly aimed at scenarios that require AI clients or agents to have internet retrieval and web content acquisition capabilities: finding relevant pages through real-time web search, then extracting page content in clean Markdown format for subsequent reading, citation, or processing.

Main capabilities The server exposes two tools: search_web_pages (search web pages) and fetch_page_content (fetch page content). Content is returned in clean Markdown format.

Usage The service is hosted, and the official recommendation is to connect directly to the remote endpoint. The hosted endpoint is https://api.keenable.ai/mcp,采用 Streamable HTTP protocol. Clients that support remote MCP, including Cursor, Cline, VS Code, and the custom connector interface of Claude Desktop, can directly enter this URL to use it. For clients or configuration paths that only accept stdio commands (for example, Claude Desktop's claude_desktop_config.json), you can use the npm wrapper @keenable/mcp, run via npx. This wrapper acts as a lightweight stdio bridge, connects to the hosted server, and forwards all tools it exposes; it does not itself contain search logic.

Default key-free access and API key Keenable requires no API key by default, with a limit of 1,000 requests per hour. To increase this rate limit, an optional API key can be passed in.

Environment variables The wrapper supports two environment variables: KEENABLE_API_KEY, an optional API key, sent as the X-API-Key request header; KEENABLE_MCP_URL, used to override the endpoint, with the default value being https://api.keenable.ai/mcp。

Use cases Suitable for users who want to obtain real-time web search results in AI clients and extract web content in clean Markdown format, especially developers using the above MCP-supporting editors or desktop clients.

Limitations Default rate limit of 1,000 requests per hour; details such as the specific search index scope, content extraction quality, number of results, language support, paid plans, and pricing are not provided in the materials. The documentation address is https://docs.keenable.ai/mcp-server,官网为 https://keenable.ai,官方 The MCP Registry identifier is ai.keenable/web-search.