paperpal

What is paperpal?

paperpal is an MCP (Model Context Protocol) server that gives large language models access to arXiv and Hugging Face papers. It helps researchers and students search, discuss, and organize literature reviews through natural language conversations with their favourite LLMs.

How to use paperpal?

Clone the repository (git clone https://github.com/jerpint/paperpal), then configure your MCP client (Claude Desktop or Cursor) by adding a server entry that uses the uv command to run paperpal.py. Restart the client to see the tools available.

Key features of paperpal

• Searches arXiv and Hugging Face papers
• Enables natural language discussion of papers
• Helps discover new research on a topic
• Aids in organising ideas for literature reviews

Use cases of paperpal

• Discuss the findings of a specific paper with an LLM
• Search for recent papers on a given topic
• Brainstorm and structure a literature review outline

FAQ from paperpal

What does paperpal do?

It provides LLMs with access to arXiv and Hugging Face papers, allowing you to search, discuss, and organise research literature through natural conversation.

How do I install paperpal?

Clone the repository locally and add a configuration entry to your MCP client (Claude Desktop or Cursor) that runs paperpal.py using the uv command.

What are the limitations of paperpal?

LLMs can still hallucinate, and semantic search is never perfect. The tool is as good as its components.

What runtime dependencies are required?

paperpal requires uv to be installed on the system to run the server script.

Which MCP clients are supported?

Claude Desktop App and Cursor are both supported, with explicit configuration examples provided in the README.