# Free Search

> Free Search is an open-source MCP server for web search and scraping. Keyless web search, page fetching and document reading that runs locally. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Open source
- **Platforms:** CLI, self-hosted, macOS, Linux, Windows
- **Repository:** https://github.com/sweetcornna/free-search-mcp
- **MCP Registry name:** io.github.sweetcornna/free-search-mcp
- **Categories:** web search and scraping
- **Listing updated:** 2026-11-15

## About

The headline tool is research(): one call searches several engines, fetches the top pages and returns a Markdown brief with sources and token estimate. Results from a default pool of keyless engines (DuckDuckGo, Bing, Yahoo! JAPAN and AnySearch) are merged with reciprocal rank fusion and de-duplicated, and extra engines join when a query is recent or written in Chinese or Korean.

Filters cover freshness, include and exclude domains, text matches and a category argument that routes to sources for papers, filings, packages or CVEs. Engines that return a CAPTCHA or login wall are benched and replaced. A local FTS5 cache lets cache_search find pages fetched earlier, and read_doc parses PDF, DOCX, XLSX, PPTX, EPUB and archives with pagination.

There is a Claude Code and Codex plugin that adds a verified-research skill, plus a proxy setting for engines that block your network. Browser-rendered engines need Chromium installed once.

## Features

- **One-call research:** Search, fetch and summarize into a Markdown brief with sources.
- **Multi-engine merge:** Parallel keyless engines combined by rank fusion and de-duplicated.
- **Document reading:** Parses PDFs, Office files, EPUB, CSV and archives with pagination.
- **Category routing:** Routes queries to paper, finance, news, software or security sources.
- **Local page cache:** Full-text search over pages already fetched.
- **Citation graph:** paper_graph walks references, citing works and retraction notices.

## Good for

- Research a question without signing up for a search API
- Compare two to five pages side by side
- Read a long PDF in pieces
- Trace a paper's citations and corrections
- Search archived pages from an earlier session

## Install

### Claude Code (Terminal)

```sh
claude mcp add free-search-mcp --env SEARCH_MCP_PROXY='<SEARCH_MCP_PROXY>' -- uvx free-search-mcp
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "free-search-mcp": {
      "command": "uvx",
      "args": [
        "free-search-mcp"
      ],
      "env": {
        "SEARCH_MCP_PROXY": "<SEARCH_MCP_PROXY>"
      }
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "free-search-mcp": {
      "command": "uvx",
      "args": [
        "free-search-mcp"
      ],
      "env": {
        "SEARCH_MCP_PROXY": "<SEARCH_MCP_PROXY>"
      }
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "free-search-mcp": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "free-search-mcp"
      ],
      "env": {
        "SEARCH_MCP_PROXY": "<SEARCH_MCP_PROXY>"
      }
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "free-search-mcp": {
      "command": "uvx",
      "args": [
        "free-search-mcp"
      ],
      "env": {
        "SEARCH_MCP_PROXY": "<SEARCH_MCP_PROXY>"
      }
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "free-search-mcp": {
      "command": "uvx",
      "args": [
        "free-search-mcp"
      ],
      "env": {
        "SEARCH_MCP_PROXY": "<SEARCH_MCP_PROXY>"
      }
    }
  }
}
```

## Tools

- `search`: Parallel multi-engine search, rank-merged and de-duplicated, with optional filters.
- `research`: Searches, fetches the top pages and returns a Markdown brief.
- `paper_graph`: Walks one paper's citation graph and retraction notices.
- `compare`: Fetches 2 to 5 URLs and returns side-by-side excerpts keyed by a question.
- `fetch`: Fetches a page or file as Markdown, parsed text or a description.
- `fetch_batch`: Fetches up to 20 URLs concurrently.
- `read_doc`: Parses PDF, DOCX, XLSX, PPTX, EPUB, CSV, code and archives with pagination.
- `extract_structured`: Pulls JSON-LD, OpenGraph, Twitter cards and microdata.
- `cache_search`: Full-text search across previously fetched pages.
- `engines`: Lists the search sources grouped by type.
- `download`: Saves a file to the cache downloads folder.

## FAQ

### Does it need an API key?

No. The README says it needs no search API key; an optional proxy and region can be set by environment variable.

### How do I install it?

Through the Claude Code or Codex plugin, or as the free-search-mcp package from PyPI run with uv.

### Where are downloads stored?

In a downloads folder under the cache directory, deleted automatically after 24 hours.

### Is Chromium required?

Only for browser-rendered engines and JavaScript-heavy pages; other features work without it.

### Is Free Search free?

Yes. Free Search is open source, so you can use and self-host it for free.

### How do I add Free Search to Claude Code?

Run this command in your terminal: claude mcp add free-search-mcp --env SEARCH_MCP_PROXY='<SEARCH_MCP_PROXY>' -- uvx free-search-mcp

### What are the best alternatives to Free Search?

The closest alternatives to Free Search are Scrapling MCP Server, Firecrawl MCP Server, SearXNG Search, NeuralVerge and AI-Powered directory.

## Alternatives

- [Scrapling MCP Server](https://ai-powered.com/mcp/scrapling-mcp-server): Scrapling MCP Server is an open-source MCP server for web search and scraping.
- [Firecrawl MCP Server](https://ai-powered.com/mcp/firecrawl-mcp-server): Firecrawl MCP Server is a freemium MCP server for web search and scraping.
- [SearXNG Search](https://ai-powered.com/mcp/searxng-search): SearXNG Search is an open-source MCP server for web search and scraping.
- [NeuralVerge](https://ai-powered.com/mcp/neuralverge): NeuralVerge is an MCP server for web search and scraping.
- [AI-Powered directory](https://ai-powered.com/mcp/ai-powered-directory): AI-Powered directory is a free MCP server for developer tools.
- [Shopify MCP Server (any store)](https://ai-powered.com/mcp/shopify-mcp-server-any-store): Shopify MCP Server (any store) is a paid MCP server for e-commerce.
- [Chanroll: YouTube Channel Video Scraper (Official API)](https://ai-powered.com/mcp/chanroll-youtube-channel-video-scraper-official-api): Chanroll: YouTube Channel Video Scraper (Official API) is a paid MCP server for design and media.
- [Vidvett: YouTube Video Details Scraper (Official API)](https://ai-powered.com/mcp/vidvett-youtube-video-details-scraper-official-api): Vidvett: YouTube Video Details Scraper (Official API) is a paid MCP server for design and media.

---
Source: https://ai-powered.com/mcp/free-search
