Scholar

You get a unified MCP server that searches arXiv and venue papers, downloads PDFs securely, and analyzes Kimi FAQs. It exposes these capabilities as reusable tools you can combine in any MCP client to streamline literature discovery, retrieval, and quick paper insights.

How to use

You access Scholar MCP by running it locally as an MCP-stdin server or by connecting to its HTTP endpoint if you deploy it remotely. Once running, you can invoke these tools to perform common tasks: - search_papers to query arXiv or venue lists and return a concise set of papers with details like title, authors, abstract, and PDF links. - download_single_paper to fetch a paper’s PDF directly into a folder you specify. - download_batch_papers to search and batch-download multiple PDFs, tracking successes and failures. - kimi_analysis to fetch Kimi-generated FAQs for a quick summary of a paper’s focus. Internally, the server hides site-specific statistics from tool results to keep responses clean.

How to install

Prerequisites: You need Node.js and npm installed on your system. Ensure you have a working internet connection for package installation and remote HTTP access if you plan to run the server in the cloud.

Step by step you can set up and run the Scholar MCP server locally:

1. Install dependencies

2. Build the project

3. Run in development mode

4. (Optional) Register with an MCP client or CLI to connect to the server

The following commands reflect the typical workflow you’ll use:

Configuration and runtime details

You can run Scholar MCP as a local stdio server or expose it via HTTP, depending on your deployment preference.

Troubleshooting and notes

If you cannot connect, verify the server is running and that your client is configured to point at the correct command (node path) and entry file (dist/index.js). If downloads fail, confirm the target directory has write permission and that the remote paper source is reachable.

Example configuration snippet

{
  "mcpServers": {
    "scholar-mcp": {
      "command": "node",
      "args": ["/absolute/path/to/scholar-mcp/dist/index.js"],
      "description": "Cool Papers search, download, and Kimi FAQ"
    }
  }
}

Available tools

search_papers

Query arXiv or venue papers with parameters for source, query, maxResults, skip, and sort; returns total and paper entries with id, title, externalUrl, pdfUrl, authors, abstract, subjects, publishTime, and relatedKeywords. Internal stats are removed from the output.

download_single_paper

Download a single paper by paperId from a specified source and folder; optionally customize the filename; returns mode, pdfUrl, filePath, fileSize, and paper metadata.

download_batch_papers

Batch search and download multiple papers based on source, query, and downloadFolder; outputs a summary including total results, successes with file paths, and failures with error messages.

kimi_analysis

Fetch Kimi-generated FAQs for a given paperId and return a list of question-answer pairs along with a total FAQ count.