Abstract
We introduce Paper2Agent, an automated framework that converts researchpapers into AI agents. Paper2Agent transforms research output from passiveartifacts into active systems that can accelerate downstream use, adoption, anddiscovery. Conventional research papers require readers to invest substantialeffort to understand and adapt a paper's code, data, and methods to their ownwork, creating barriers to dissemination and reuse. Paper2Agent addresses thischallenge by automatically converting a paper into an AI agent that acts as aknowledgeable research assistant. It systematically analyzes the paper and theassociated codebase using multiple agents to construct a Model Context Protocol(MCP) server, then iteratively generates and runs tests to refine and robustifythe resulting MCP. These paper MCPs can then be flexibly connected to a chatagent (e.g. Claude Code) to carry out complex scientific queries throughnatural language while invoking tools and workflows from the original paper. Wedemonstrate Paper2Agent's effectiveness in creating reliable and capable paperagents through in-depth case studies. Paper2Agent created an agent thatleverages AlphaGenome to interpret genomic variants and agents based on ScanPyand TISSUE to carry out single-cell and spatial transcriptomics analyses. Wevalidate that these paper agents can reproduce the original paper's results andcan correctly carry out novel user queries. By turning static papers intodynamic, interactive AI agents, Paper2Agent introduces a new paradigm forknowledge dissemination and a foundation for the collaborative ecosystem of AIco-scientists.