ArXiv Scout — MCP server for Claude Code
ArXiv Scout searches arXiv, extracts text from PDFs, and tracks citations through Semantic Scholar. No API key required; gather research materials and build literature reviews in seconds.
MIT 3,426 installs ★ 1 updated 2026-06-21
claude mcp add --transport http arxiv-scout https://arxiv-scout--shawnnygoh.run.toolsWhat ArXiv Scout does
ArXiv Scout connects to any MCP-compatible IDE and exposes eight tools for academic research. Search with category and date filters, fetch metadata by ID, download PDFs with automatic text extraction and section detection, browse outbound and inbound citations. Built-in caching speeds up repeated lookups.
Three bundled prompts handle common workflows: structured paper summaries (problem, methods, results, limitations), side-by-side comparison of up to five papers, and synthesis of literature reviews on demand. The server pulls from arXiv and Semantic Scholar APIs, so you only need internet access; hosting on Smithery means no local installation.
For local setup, you'll need Python and uv. Configuration is straightforward: port and host via environment variables; optional Semantic Scholar API key lifts rate limits. Development includes unit tests and integration test modes.
What it helps with
- find recent papers in a narrow field filtered by date and category
- extract and summarize text from academic PDF files
- compare methodologies and results across multiple studies
- trace citation chains and discover related research
- compile materials for a literature review on any topic
Example prompts for the agent
- Find the three most recent papers on transformers published this month
- Download and parse the PDF from arxiv:2401.12345, extract major sections
- Compare these two papers side by side: methods, findings, and gaps
How to install ArXiv Scout
Platform CLI: remote server
claude mcp add --transport http arxiv-scout https://arxiv-scout--shawnnygoh.run.toolsPlatform CLI: remote server
codex mcp add arxiv-scout --url https://arxiv-scout--shawnnygoh.run.toolsConfig file
{
"mcpServers": {
"arxiv-scout": {
"url": "https://arxiv-scout--shawnnygoh.run.tools"
}
}
}Config file
{
"servers": {
"arxiv-scout": {
"type": "http",
"url": "https://arxiv-scout--shawnnygoh.run.tools"
}
}
}Platform CLI: remote server
gemini mcp add -t http arxiv-scout https://arxiv-scout--shawnnygoh.run.toolsStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about ArXiv Scout
How do I install ArXiv Scout in Claude Code?
Run claude mcp add --transport http arxiv-scout https://arxiv-scout--shawnnygoh.run.tools. The server connects immediately; no local code needed.
Do I need an API key?
No. Search and PDF retrieval work free through open arXiv and Semantic Scholar. An optional Semantic Scholar key only helps if you hit rate limits.
Does it work in other IDEs besides Claude Code?
Yes: Codex, Cursor, GitHub Copilot, Windsurf, Cline, Zed, and Gemini CLI. Each has its own config format in the installation instructions.