ElevenLabs MCP Server

Instructions

Args:
    page: Page number to return (0-indexed)
    page_size: Number of voices to return per page (1-100)
    search: Search term to filter voices by

Returns:
    TextContent containing information about the shared voices

Implementation Reference

• elevenlabs_mcp/server.py:1010-1066 (handler)

  The handler function implementing the 'search_voice_library' tool. It fetches shared voices from ElevenLabs using client.voices.get_shared(), processes voice details including languages, gender, age, etc., and returns formatted TextContent.

  def search_voice_library(
      page: int = 0,
      page_size: int = 10,
      search: str | None = None,
  ) -> TextContent:
      response = client.voices.get_shared(
          page=page,
          page_size=page_size,
          search=search,
      )
  
      if not response.voices:
          return TextContent(
              type="text", text="No shared voices found with the specified criteria."
          )
  
      voice_list = []
      for voice in response.voices:
          language_info = "N/A"
          if hasattr(voice, "verified_languages") and voice.verified_languages:
              languages = []
              for lang in voice.verified_languages:
                  accent_info = (
                      f" ({lang.accent})"
                      if hasattr(lang, "accent") and lang.accent
                      else ""
                  )
                  languages.append(f"{lang.language}{accent_info}")
              language_info = ", ".join(languages)
  
          details = [
              f"Name: {voice.name}",
              f"ID: {voice.voice_id}",
              f"Category: {getattr(voice, 'category', 'N/A')}",
          ]
          # TODO: Make cleaner
          if hasattr(voice, "gender") and voice.gender:
              details.append(f"Gender: {voice.gender}")
          if hasattr(voice, "age") and voice.age:
              details.append(f"Age: {voice.age}")
          if hasattr(voice, "accent") and voice.accent:
              details.append(f"Accent: {voice.accent}")
          if hasattr(voice, "description") and voice.description:
              details.append(f"Description: {voice.description}")
          if hasattr(voice, "use_case") and voice.use_case:
              details.append(f"Use Case: {voice.use_case}")
  
          details.append(f"Languages: {language_info}")
  
          if hasattr(voice, "preview_url") and voice.preview_url:
              details.append(f"Preview URL: {voice.preview_url}")
  
          voice_info = "\n".join(details)
          voice_list.append(voice_info)
  
      formatted_info = "\n\n".join(voice_list)
      return TextContent(type="text", text=f"Shared Voices:\n\n{formatted_info}")

• elevenlabs_mcp/server.py:998-1009 (registration)

  The @mcp.tool decorator registers the 'search_voice_library' tool with MCP, including the description that serves as input/output schema.

  @mcp.tool(
      description="""Search for a voice across the entire ElevenLabs voice library.
  
      Args:
          page: Page number to return (0-indexed)
          page_size: Number of voices to return per page (1-100)
          search: Search term to filter voices by
  
      Returns:
          TextContent containing information about the shared voices
      """
  )

• elevenlabs_mcp/server.py:999-1009 (schema)

  The tool description defines the input parameters (page, page_size, search) and return type (TextContent), serving as the schema.

      description="""Search for a voice across the entire ElevenLabs voice library.
  
      Args:
          page: Page number to return (0-indexed)
          page_size: Number of voices to return per page (1-100)
          search: Search term to filter voices by
  
      Returns:
          TextContent containing information about the shared voices
      """
  )