Librivox

What the Librivox MCP server does

The Librivox MCP server connects AI agents to the LibriVox public-domain audiobook catalog, which contains roughly 17,000 titles in multiple languages. It focuses on catalog discovery and metadata rather than audio playback. The underlying data source is the LibriVox feed at https://librivox.org/api/feed/.

The Librivox MCP server can search audiobook records by general query, author, title, or a date-related since parameter. Results support pagination through limit and offset. Separate author search supports a general query, last name, and pagination. Once an audiobook identifier is known, an agent can retrieve that audiobook’s details and request its track list.

Tools and capabilities

The pack provides four primary tools:

• audiobooks(query?, author?, title?, since?, limit?, offset?) searches the audiobook catalog.
• audiobook(id) retrieves one audiobook by identifier.
• authors(query?, last_name?, limit?, offset?) searches authors.
• tracks(project_id) lists the tracks associated with an audiobook.

These tools are suitable for locating public-domain books, narrowing results by author or title, inspecting a selected audiobook, and obtaining its chapter or track entries.

How it works

The Librivox MCP server supports two connection modes. Through Pipeworx, an MCP client connects to https://gateway.pipeworx.io/librivox/mcp. That scoped gateway endpoint lists the four LibriVox tools along with shared Pipeworx meta-tools. The gateway’s ask_pipeworx tool can route natural-language questions across the broader Pipeworx catalog, including sources beyond LibriVox.

A full Pipeworx connection at https://gateway.pipeworx.io/mcp exposes the broader catalog directly. Both gateway URLs use the same gateway and data sources; the difference is which pack’s tools appear directly in the tool list.

The package also runs as a standalone local MCP process. In that mode, it communicates over standard input and output and exposes only the LibriVox pack’s tools. It does not include the gateway’s shared meta-tools or cross-catalog routing.

Setup and configuration

For a gateway connection, add the LibriVox endpoint to an MCP client configuration:

{
  "mcpServers": {
    "librivox": {
      "url": "https://gateway.pipeworx.io/librivox/mcp"
    }
  }
}

The README names Claude Desktop, Cursor, and Windsurf as client options.

For local use, configure the package with npx:

{
  "mcpServers": {
    "librivox": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-librivox"]
    }
  }
}

You can also verify that the local process starts with npx -y @pipeworx/mcp-librivox.

Limitations and notes

The gateway endpoint includes approximately 30 shared tools in addition to the pack-specific tools, so its tool list is larger than the four-tool catalog above. The exact scope is stated by the MCP initialize response and may vary. There is currently no gateway option to remove those shared meta-tools from a pack endpoint.

For clients that do not use MCP, the gateway offers HTTP tool calls. For example, audiobook search can be sent as a POST request to /v1/tools/audiobooks with a JSON body such as {\"query\":\"Pride and Prejudice\"}. Tool metadata can be inspected with a GET request to the corresponding tool endpoint.

The local Librivox MCP server avoids the gateway round trip and account requirements, but it cannot use ask_pipeworx to route questions to other data packs.