academia_mcp

Academia MCP is a Python-based MCP server that provides tools to search, fetch, analyze, and report on scientific papers and datasets. It exposes a set of built-in capabilities for querying sources such as arXiv, ACL Anthology, Hugging Face datasets, Semantic Scholar, and generic web search. The server also supports LaTeX workflow helpers, PDF reading, and optional large language model (LLM) powered QA and proposal generation. You can run the server over HTTP (default) or use stdio for local MCP clients. The included tools allow you to perform targeted searches (arXiv, Anthology, Hugging Face datasets), retrieve and read papers (PDF/text, LaTeX templates, and PDF extraction), and generate structured reports or research proposals with optional LLM assistance. To interact, start the server with a transport, for example streaming HTTP, and then call the available tools through the MCP client interface.

Prerequisites:

• Python 3.12+ (recommended) and pip
• Optional: uv for development workflow (uv run -m academia_mcp ...)

Install from PyPI (end users):

1. Ensure Python is installed and available as python3
2. Install the package:

pip3 install academia-mcp

Development / local testing (uv + Makefile):

1. Create a virtual environment and install dev dependencies:

uv venv .venv
make install

Run the server (HTTP by default):

# Using Python module directly
python -m academia_mcp --transport streamable-http

# Or using uv (for development workflow)
uv run -m academia_mcp --transport streamable-http

Tips:

• If you plan to enable authentication, set ENABLE_AUTH=true and optionally TOKENS_FILE to point to a tokens.json file.
• For HTTP transport, PORT can be customized via the PORT environment variable (default 5056).
• Explore the available tools via the CLI (e.g., arxiv_search, s2_get_citations, read_pdf, compile_latex).

Environment variables and configuration options:

• ENABLE_AUTH: Enable token-based authentication for HTTP transports (default: false)
• TOKENS_FILE: Path to tokens.json (default: ./tokens.json)
• OPENROUTER_API_KEY, EXA_API_KEY, BRAVE_API_KEY, TAVILY_API_KEY: Enable respective web/search providers for web_search and visit_webpage features
• WORKSPACE_DIR: Directory for generated files (PDFs, temp artifacts) when using LaTeX/PDF tools
• PORT: HTTP port (default 5056)

Docker users can build and run the image and pass environment variables to customize behavior, as shown in the Docker example in the README. If you encounter issues with LaTeX or PDF tooling, ensure LaTeX is installed and on PATH (pdflatex and latexmk).