Stanford's Paper2Agent turns a research paper and its code into an MCP server any agent can run
Published in Nature, the MIT-licensed pipeline builds a 'virtual corresponding author' that reproduces results and applies methods to new data through Claude Code or Codex.
A Stanford team led by Jiacheng Miao and James Zou has published Paper2Agent in Nature, dated 16 September 2026. The system converts a computational research paper and its codebase into a Model Context Protocol (MCP) server, so any MCP-compatible agent such as Claude Code can run the paper's methods through natural language. The authors describe the result as a 'virtual corresponding author.' The motivation is practical. Computational papers ship code that readers must clone, install, configure and debug, which keeps useful methods locked inside PDFs. Paper2Agent runs on Claude Code's agent SDK. A central orchestrator dispatches specialized sub-agents through six steps that begin with locating and downloading the codebase, building an isolated virtual environment, indexing usable tutorials, and executing those tutorials end to end while recording reference outputs. The finished server can reproduce the paper's results or run its methods on new data. The code is MIT-licensed and installs as a skill for Claude Code or Codex. Prebuilt servers for AlphaGenome, Scanpy and TISSUE run on Hugging Face Spaces, and a hosted version is also available. For researchers, the approach could shorten the gap between reading a method and applying it, and it lowers the cost of checking whether a published result reproduces. It also illustrates a broader pattern in which MCP becomes the packaging format for tools of many kinds, from scientific code to the Google Home server released the same day.