A collection of MCP tools related to the search of scientific papers:

• ArXiv search and download
• ACL Anthology search
• HuggingFact datasets search
• Semantic Scholar citation graphs
• Web search: Exa/Brave/Tavily
• Page crawler

• Using pip (end users):

pip3 install academia-mcp

• For development (uv + Makefile):

uv venv .venv
make install

Comprehensive report screencast: https://www.youtube.com/watch?v=4bweqQcN6w8

Single paper screencast: https://www.youtube.com/watch?v=IAAPMptJ5k8

{
  "mcpServers": {
    "academia": {
      "command": "python3",
      "args": [
        "-m",
        "academia_mcp",
        "--transport",
        "stdio"
      ]
    }
  }
}

uv run -m academia_mcp --transport streamable-http

Notes:

• Transports supported: stdio, sse, streamable-http.
• Host/port are used for HTTP transports; for stdio they are ignored.

• make install: install the package in editable mode with uv.
• make validate: run black, flake8, and mypy (strict).
• make test: run the test suite with pytest.
• make publish: build and publish using uv.

Set as needed depending on which tools you use:

• TAVILY_API_KEY: enables Tavily in web_search.
• EXA_API_KEY: enables Exa in web_search and visit_webpage.
• BRAVE_API_KEY: enables Brave in web_search.
• OPENROUTER_API_KEY: required for document_qa.
• BASE_URL: override OpenRouter base URL for document_qa and bitflip tools.
• DOCUMENT_QA_MODEL_NAME: override default model for document_qa.
• BITFLIP_MODEL_NAME: override default model for bitflip tools.
• WORKSPACE_DIR: directory for generated files (PDFs, temp artifacts).

The md_to_pdf tool invokes pdflatex. Ensure a LaTeX distribution is installed and pdflatex is on PATH. On Debian/Ubuntu:

sudo apt install texlive-latex-base texlive-fonts-recommended texlive-latex-extra texlive-science