Skip to main content
AdOpenFree logoPromote your productReach more potential users and drive product growth and revenue.Advertise

OpenFree MCP Guide

Use the Model Context Protocol (MCP) to search OpenFree’s directory of free and open-source AI tools from your AI assistant.

Server URL
https://openfree.tools/mcp
Transport
Streamable HTTP
Authentication
No account or API key required

Connect your client

Claude Code

claude mcp add --transport http openfree https://openfree.tools/mcp

Codex

codex mcp add openfree --url https://openfree.tools/mcp

Other MCP clients

Add a remote server named openfree, select Streamable HTTP, and paste the server URL above. If your client accepts mcpServers JSON with HTTP transport, use this configuration.

{
  "mcpServers": {
    "openfree": {
      "type": "http",
      "url": "https://openfree.tools/mcp"
    }
  }
}

Try your first search

Open a new conversation in your connected client and try this prompt.

Use OpenFree to search for Ollama. Return three results with their GitHub repositories and OpenFree links.

Available tools

All four tools provide read-only access to published directory data. Set locale to zh for Chinese results or en for English (the default).

search_tools
Search by keyword (q). Filter by category or tag slug and paginate with page and limit. The default limit is 10 and the maximum is 25.
get_tool
Get details for a published tool by its exact slug, such as ollama.
list_categories
List categories with published tools, including their slugs and tool counts.
list_tags
List the 50 most-used tags on published tools, including their slugs and tool counts.

Call the API over HTTP

Send JSON-RPC POST requests with Content-Type set to application/json and Accept set to application/json, text/event-stream. Send initialize, then notifications/initialized. Use tools/list to discover the input schemas. Send the negotiated MCP-Protocol-Version on subsequent requests, including the tools/call example below. No session ID is required.

{
  "jsonrpc": "2.0",
  "id": 3,
  "method": "tools/call",
  "params": {
    "name": "search_tools",
    "arguments": {
      "q": "Ollama",
      "locale": "zh",
      "limit": 3
    }
  }
}

The tool returns JSON text in result.content[0].text. Opening the server URL directly in a browser sends a GET request and returns 405; this is expected.