mcp-pandoc is a Model Context Protocol (MCP) server that enables document format conversion using Pandoc. It provides tooling to transform content between common formats while preserving formatting and structure, leveraging the Pandoc Python package as the conversion engine. The server is in early development; PDF support is under development and capabilities may evolve. Core functionality is exposed via the convert-contents tool, which accepts either raw contents or a path to an input file, and converts it to a target format. Options include input_format (default markdown), output_format (default markdown), and output_file (required for advanced formats such as pdf, docx, rst, latex, epub). Additional features include a reference_doc to style DOCX output and a defaults_file (YAML) to define reusable templates, plus optional pandoc filters. Supported formats include markdown, html, pdf, docx, rst, latex, epub, txt, ipynb, and odt, with bidirectional conversions among them. Prerequisites include Pandoc, the UV package, and, for PDF, TeX Live. Debugging can be aided by MCP Inspector, and the project aims to stay extensible within the MCP ecosystem.