Academia

Academia MCP is a server that lets you search, fetch, analyze, and report on scientific papers and datasets from multiple sources. It combines source discovery with optional AI-assisted workflows, so you can build research proposals, read PDFs, and extract insights from literature in a single, programmable interface.

How to use

Run Academia MCP locally in stdio mode if you want tight integration with local MCP clients (for example, a desktop app). Use HTTP transport if you need to connect over a network. If you enable authentication, you can issue and manage tokens to control access.

Starting the server in stdio mode (local use): run the following command to start the MCP in stdio transport so a client like Claude Desktop can communicate directly on your machine.

python3 -m academia_mcp --transport stdio

Authentication and access control

Authentication is optional for HTTP transports and can be enabled to require tokens in requests. Tokens are issued and revoked through dedicated commands. When enabled, you’ll pass the token in the Authorization header as a Bearer token or as a query parameter apiKey.

Enable authentication in your environment to activate token checks.

Example environment setup for authentication (adjust paths as needed):

export ENABLE_AUTH=true
export TOKENS_FILE=/path/to/tokens.json  # optional, defaults to ./tokens.json

Using environment variables and workspace

Certain features require environment configuration. For example, enabling AI-powered document QA or research-proposal tools requires an AI service API key, and working with LaTeX/PDF features requires a workspace directory.

Configure the following environment variables as needed:

OPENROUTER_API_KEY: enables LLM-related tools
BASE_URL: override OpenRouter base URL
DOCUMENT_QA_MODEL_NAME: override model for document_qa
BITFLIP_MODEL_NAME: override model for bitflip tools
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
WORKSPACE_DIR: directory for generated files (PDFs, temp artifacts)
PORT: HTTP port (default 5056)

Typical data sources and tools you can use

Academia MCP provides a set of tools to search and process academic content, including arXiv, ACL Anthology, Hugging Face datasets, Semantic Scholar, and web sources. You can also compile LaTeX documents, read PDFs, and run AI-assisted analysis workflows.

Available tool families range from searching and downloading papers to reading web pages, handling LaTeX templates, and performing document QA or proposal generation with AI.

Available tools

arxiv_search

Query arXiv with field-specific queries and filters.

arxiv_download

Fetch a paper by ID and convert to structured text (HTML/PDF modes).

anthology_search

Search ACL Anthology with fielded queries and optional date filtering.

hf_datasets_search

Find Hugging Face datasets with filters and sorting.

s2_get_citations

List papers citing a given arXiv paper (Semantic Scholar Graph).

s2_get_references

List papers referenced by a given arXiv paper.

visit_webpage

Fetch and normalize a web page.

web_search

Unified search wrapper; available when at least one provider key is set.

exa_web_search

Provider-specific search with Exa.

brave_web_search

Provider-specific search with Brave.

tavily_web_search

Provider-specific search with Tavily.

get_latex_templates_list

Enumerate built-in LaTeX templates.

get_latex_template

Fetch a built-in LaTeX template.

compile_latex

Compile LaTeX to PDF in WORKSPACE_DIR.

read_pdf

Extract text per page from a PDF.

download_pdf_paper

Download a paper PDF.

review_pdf_paper

Review PDFs (requires LLM + workspace).

document_qa

Answer questions over provided document chunks (requires LLM).

extract_bitflip_info

Extract information for research proposals using LLM.

generate_research_proposals

Generate draft research proposals (LLM).

score_research_proposals

Score and compare proposed research ideas (LLM).