mcp-dblp is a Model Context Protocol (MCP) server that provides access to the DBLP computer science bibliography database for Large Language Models. It bridges the DBLP API with LLM workflows, enabling models to search and retrieve publications, manage citations, and format bibliographic information. It can perform boolean search queries, fuzzy matching for titles and author names, and returns BibTeX entries directly from DBLP. The server supports filtering publications by year and venue, computes basic statistics, and processes embedded references within documents. A key capability is direct BibTeX export that bypasses LLM processing to ensure accuracy. The MCP server exposes a set of tools for common tasks, including get_instructions, search, fuzzy_title_search, get_author_publications, get_venue_info, calculate_statistics, add_bibtex_entry, and export_bibtex, facilitating flexible workflows. It is designed to run on Python 3.11+ and uv, and emphasizes accuracy and direct DBLP data during BibTeX export to avoid hallucination.