PDF Reader MCP Server

Instructions

Implementation Reference

• src/server/main.py:64-83 (handler)

  The handler function for the 'get_pdf_metadata' tool, registered via @mcp.tool(). It loads the PDF using PDFParser.loader, extracts metadata (page_count, title, author), converts to string, and returns it.

  @mcp.tool()
  async def get_pdf_metadata(source: str) -> str:
      """
      Quickly retrieve metadata from a PDF without reading the full content.
      """
      # We can reuse parser but limit the logical scope.
      # Actually parser loads the doc anyway.
      # For optimization, we might make a separate method in parser, but for now reuse.
  
      # We load it but generate result without full extraction
      doc = await parser.loader.load(source)
      try:
          meta = {
              "page_count": len(doc),
              "title": doc.metadata.get("title", ""),
              "author": doc.metadata.get("author", ""),
          }
          return str(meta)
      finally:
          doc.close()