arxiv-smart-mcp
Allows searching and reading arXiv papers, including fetching metadata, downloading PDFs, and retrieving HTML or markdown conversions.
This server does not currently have a deployable release on Glama. Follow the repository's setup instructions to install and configure it for your MCP client, including any required credentials and local dependencies.
arXiv MCP Server and Rate Limiting Proxy
MCP server for searching and reading arXiv papers from AI assistants — self-rate-limited to avoid getting blocked by the arXiv API.
Why not a simple MCP server?
The arXiv API enforces strict rate limits. If you exceed them, your IP gets blocked. A naive MCP server that calls the arXiv API directly has no way to throttle requests — and AI agents tend to fire many requests in quick succession.
This project solves that by routing all requests through a local FastAPI proxy that enforces rate limiting automatically. Your AI assistant can call as many tools as it wants without getting banned.
AI assistant → MCP server → FastAPI proxy (rate-limited) → arXiv APIRelated MCP server: Research Paper Agent
Tools
When connected, your AI agent gets access to these tools:
search_papers — search arXiv by query with sorting options
get_paper — get full metadata for a paper by arXiv ID
download_pdf — download a paper's PDF
get_paper_html — get the HTML rendering from ar5iv
get_paper_markdown — get a markdown conversion of the paper
Setup
Prerequisites: Python 3.12+, uv, just
1. Initialize and start the proxy
just init
just startThe proxy must be running for the MCP tools to work. The MCP server connects to the proxy on http://127.0.0.1:7171 by default.
2. Connect the MCP server
Paste this to your AI assistant and ask it to add this MCP server:
{
"mcpServers": {
"arxiv-smart": {
"command": "npx",
"args": ["tsx", "mcp/arxiv-smart.ts"],
"cwd": "/path/to/arxiv-smart-mcp"
}
}
}Replace /path/to/arxiv-smart-mcp with the actual path to this repository.
Configuration
The proxy is configured via config.yaml, which is created from config.yaml.template during just init. Key settings:
port — proxy listen port (default: 7171)
rate_limit_seconds — minimum interval between arXiv API calls (default: 3.0)
request_timeout_seconds — timeout for arXiv API requests (default: 30.0)
If you change the port, set the REST_BASE environment variable in your MCP config so the MCP server can find the proxy:
{
"mcpServers": {
"arxiv-smart": {
"command": "npx",
"args": ["tsx", "mcp/arxiv-smart.ts"],
"cwd": "/path/to/arxiv-smart-mcp",
"env": {
"REST_BASE": "http://127.0.0.1:7171"
}
}
}
}License
This server cannot be deployed
Maintenance
Related MCP Connectors
Driflyte MCP server which lets AI assistants query topic-specific knowledge from web and GitHub.
Academic research MCP server for paper search, citation checks, graphs, and deep research.
MCP server that lets your AI assistant build and deploy apps, hosting each one at a live URL.
Related MCP Servers
- AlicenseAqualityFmaintenanceAn MCP server that provides access to arXiv papers through their API.41,525 PyPI202MIT
- FlicenseNot gradedqualityDmaintenanceA remote MCP server for searching arXiv papers, extracting paper details, and generating structured prompts for LLM agents.1-
- FlicenseAqualityDmaintenanceA streamlined MCP server that connects AI assistants to arXiv's vast collection of academic papers, enabling search, retrieval, and analysis of research papers.71-
- FlicenseAqualityDmaintenanceA local MCP server for searching and reading arXiv papers, enabling paper search, retrieval, and summarization through Claude.6-