Stanford's Paper2Agent Converts Research Repos into Executable AI Assistants

A Stanford University research team has developed Paper2Agent, a system that transforms academic research repositories on GitHub into interactive AI agents capable of running actual code. The tool clones a research repository, executes tutorial files, extracts functional components, and assembles them into a standardized tool server that users can query directly. Published in Nature on September 16, the system was tested on 100 bioinformatics papers from bioRxiv, successfully converting 74, while achieving roughly 98% accuracy on a separate set of 10 non-biology repositories. Failures were attributed to incomplete code, poor documentation, or unresolvable dependencies — limitations the team disclosed transparently. The project is open-source under an MIT license but requires an external AI agent such as Claude or Gemini CLI to operate, with API costs estimated at around $14–15 per complex repository and processing times ranging from 45 minutes to three hours.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Discussion (0)
Log in to join the discussion and vote.
Log in