anysearch-ai/anysearch-mcp-server

What it solves

AnySearch MCP Server provides a unified interface for AI agents to perform real-time web searches and content extraction. It eliminates the need for multiple search APIs by offering general web search, specialized vertical domain searches (such as finance, academic, and legal), and the ability to extract full-page content as Markdown for better LLM consumption.

How it works

The project is an MCP (Model Context Protocol) server that connects AI agents (like Claude Desktop, Cursor, or OpenCode) to the AnySearch API. It supports multiple transport methods, including native Streamable HTTP, SSE, and stdio (via proxies like mcp-remote or supergateway).

Agents can use the following tools:

• search: Performs general or domain-specific searches.
• get_sub_domains: Retrieves the available vertical domains and their required parameter schemas.
• batch_search: Executes up to five independent search queries in parallel.
• extract: Converts a target URL's HTML content into Markdown.

Who it’s for

Developers and users of AI agents that support the MCP standard, who want to give their agents the ability to browse the web, research specialized domains, and extract clean content from websites.

Highlights

• Unified Search: Combines general web search with vertical domain search in one tool.
• extract Tool: Fetches full-page content and returns it as Markdown.
• Parallel Processing: Supports batch searching to execute multiple queries simultaneously.
• Flexible Transport: Native support for Streamable HTTP, with proxy support for SSE and stdio.
• Anonymous Access: Works without an API key, though with lower rate limits.