Data as of

ArXiv MCP Server for Cursor: Setup & Config

Cursor: Search arXiv, download papers, and read their contents through an AI assistant.

What you can do with ArXiv in Cursor

A focused arXiv workflow from filtered discovery to locally stored, readable papers.

  • Read long papers in bounded Markdown sections.
  • Use guided prompts for comparisons or literature reviews.

Before you start: ArXiv in Cursor

  • Python 3.11+
  • uv
  • An MCP-compatible client

Replace the example storage path with your paper directory. Install the [pdf] extra for PDF-only papers. The README also documents uvx and HTTP launch options.

ArXiv Cursor configuration

  1. Choose Cursor configuration scope

    For ArXiv MCP Server, open Project: .cursor/mcp.json User: ~/.cursor/mcp.json. Merge into mcpServers at project or user scope.

  2. Add ArXiv MCP Server to Cursor

    Use the arxiv-mcp-server entry below in mcpServers. Keep ArXiv MCP Server's documented stdio connection fields and your other Cursor entries.

  3. Check ArXiv MCP Server in Cursor

    ArXiv MCP Server: Enable in Cursor’s MCP settings; inspect tools.

Config file · Project: .cursor/mcp.json

User: ~/.cursor/mcp.json

JSON
{
  "mcpServers": {
    "arxiv-mcp-server": {
      "command": "uv",
      "args": [
        "tool",
        "run",
        "arxiv-mcp-server",
        "--storage-path",
        "/path/to/paper/storage"
      ],
      "env": {}
    }
  }
}

ArXiv tools available to Cursor

ArXiv's stored tool list documents 5 tools; compare it with Cursor's connected tool list, because the stored list is not a live check.

search_papers

Find search_papers under ArXiv in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Find arXiv papers using topic queries and optional date or category filters.

download_paper

Find download_paper under ArXiv in Cursor's MCP settings; inspect its arguments when the agent requests a call.

list_papers

Find list_papers under ArXiv in Cursor's MCP settings; inspect its arguments when the agent requests a call.

read_paper

Find read_paper under ArXiv in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Read a saved paper as Markdown, optionally in bounded chunks.

semantic_search

Find semantic_search under ArXiv in Cursor's MCP settings; inspect its arguments when the agent requests a call.

Search downloaded papers by semantic similarity; this feature is experimental.

ArXiv: full stored tool descriptions

Troubleshooting ArXiv in Cursor

  • Cursor must find uv for ArXiv MCP Server. Enable the project/user entry in MCP settings and inspect its tools.
  • ArXiv MCP Server uses stdio here: Cursor must keep the configured process running; protocol output belongs on stdout and diagnostics on stderr.
  • Known limitation: PDF-only papers need the additional PDF dependency extra.
  • Known limitation: Local semantic search is experimental and requires extra dependencies.
  • Known limitation: Untrusted paper text can carry prompt-injection instructions.
  • General Cursor troubleshooting lives in the /clients/cursor guide.

ArXiv MCP Server — details. Cursor guide

Frequently asked questions

Cursor's ArXiv MCP Server entry belongs under mcpServers in .cursor/mcp.json (project) or ~/.cursor/mcp.json (user). Confirm the active scope in MCP settings.

Cursor's arxiv-mcp-server example uses stdio.

ArXiv MCP Server's Cursor configuration invokes uv.

ArXiv MCP Server's stored list contains 5 tools, including search_papers, download_paper, list_papers. Compare those names with Cursor's connected tool list; the stored list is not a live check.

No. The README identifies the supported distribution as a Python package and warns about an unrelated npm package.

Yes. Streamable HTTP is documented alongside the default stdio transport, including loopback binding and proxy security guidance.