Academic

You can run Academic MCP to search, download, and read academic papers from multiple sources through any MCP-compatible client. It provides three core tools—paper_search, paper_download, and paper_read—so you can find papers, retrieve PDFs, and extract text content in a unified workflow.

How to use

Install the Academic MCP server locally and run it as an MCP endpoint. Connect your MCP client to the local server to perform three main actions: search for papers across supported databases, download the requested PDFs, and read or extract text from those papers. Use the client to submit search queries, download results by paper identifier, and then read the text content for quick review or context expansion.

How to install

Prerequisites you need before installing: Python 3.x and pip. You also need an MCP-compatible client to connect to the server after it is running.

Step 1. Install the Academic MCP package.

pip install academic-mcp

Step 2. Prepare the MCP client configuration. The following configuration enables the local, stdio-based server and sets optional API keys and download paths.

{
  "mcpServers": {
    "academic_mcp": {
      "command": "python",
      "args": [
        "-m",
        "academic_mcp"
      ],
      "env": {
        "SEMANTIC_SCHOLAR_API_KEY": "",
        "ACADEMIC_MCP_DOWNLOAD_PATH": "./downloads"
      }
    }
  }
}

Step 3. Start the server. You can run the server directly from the command line by invoking Python with the module name shown in the configuration.

SEMANTIC_SCHOLAR_API_KEY=YOUR_API_KEY ACADEMIC_MCP_DOWNLOAD_PATH=./downloads python -m academic_mcp

Additional configuration notes

You can customize environment variables to enable enhanced features. The SEMANTIC_SCHOLAR_API_KEY is optional and enables extended Semantic Scholar capabilities, while ACADEMIC_MCP_DOWNLOAD_PATH controls where downloaded PDFs are saved.

Environment variables

The following environment variables are used by the local server configuration:

• SEMANTIC_SCHOLAR_API_KEY: Optional API key for enhanced Semantic Scholar features (empty by default).

• ACADEMIC_MCP_DOWNLOAD_PATH: Directory for downloaded PDFs (default: ./downloads).

Security and maintenance

Keep your API keys secret and avoid exposing the local server to untrusted networks. Regularly update the Academic MCP package to benefit from security fixes and new features. If you use a public MCP client, consider restricting access to the local server and using authentication where available.

Troubleshooting

If the server does not start, verify that Python is installed and the package is correctly installed. Check that the environment variables are set as shown and that the download path exists and is writable.

Examples of usage with the MCP client

# Search papers
paper_search([
  {"searcher": "arxiv", "query": "machine learning", "max_results": 5}
])

# Download papers
paper_download([
  {"searcher": "arxiv", "paper_id": "2106.12345"}
])

# Read papers
paper_read(searcher="arxiv", paper_id="2106.12345")

Notes for developers

This server exposes three main tools through an MCP interface: paper_search, paper_download, and paper_read. It is designed to be extended with additional academic platforms by updating the sources module, while keeping the same unified interface for clients.

Available tools

paper_search

Search papers across multiple academic platforms using unified query parameters.

paper_download

Download paper PDFs by identifiers and return file paths to the downloaded files.

paper_read

Extract and read text content from downloaded papers for quick review.