- Home
- MCP servers
- ArXiv Paper
ArXiv Paper
- javascript
17
GitHub Stars
javascript
Language
6 months ago
First Indexed
2 months ago
Catalog Refreshed
Documentation & install
Readme and setup notes from the catalogue, plus a client-ready config you can copy for your MCP host.
Installation
Add the following to your MCP client configuration file.
Configuration
View docs{
"mcpServers": {
"yzfly-arxiv-paper-mcp": {
"command": "npx",
"args": [
"-y",
"@langgpt/arxiv-paper-mcp@latest"
],
"env": {
"DEBUG": "arxiv-paper-mcp"
}
}
}
}You can run and use the ArXiv Paper MCP server to search for arXiv papers, fetch direct PDF links, and parse paper content through MCP-compatible clients. This server follows the Model Context Protocol (MCP), letting you interact with it using standard MCP tooling and workflows.
How to use
Set up an MCP client to talk to the ArXiv Paper MCP server. You can run the server locally via MCP-compliant clients or use a remote MCP endpoint if available. Typical workflows let you search for papers, retrieve direct PDF URLs, and parse paper content with optional metadata to enrich results.
How to install
# NPX: quick start without installation
npx @langgpt/arxiv-paper-mcp
# Global installation (CLI)
npm install -g @langgpt/arxiv-paper-mcp
arxiv-paper-mcp
# Alternatively, for MCP client configuration (see below)
Configuration and usage notes
To connect an MCP client, you can use a local, stdio-based setup that runs the MCP server command through npx. This is particularly compatible with Claude Desktop and other MCP clients that expect a local process. The configuration example below shows how to wire the MCP server so your client can start and communicate with it.
Troubleshooting and tips
If you encounter difficulties, verify you can run the MCP server command locally and that your MCP client is configured to use stdio transport. For debugging, enable verbose logging as shown in the usage example to surface any startup or runtime errors.
Notes on security and environment
Run the MCP server with local tooling you trust, and avoid exposing sensitive endpoints publicly. If you enable verbose debugging, ensure logs do not leak credentials. Use environment controls to limit access to your machine or network as appropriate.
Available tools
search_arxiv
Search arXiv papers by keywords and retrieve a list of matching results.
get_arxiv_pdf_url
Resolve the direct PDF download URL for a given arXiv paper (by URL or arXiv ID).
parse_paper_content
Parse and extract paper content, prioritizing HTML and falling back to PDF when needed.
get_recent_ai_papers
Fetch today’s latest AI-focused papers from arXiv.