Scholar MCP — Claude Code MCP server: what it does, how to install
Searches academic papers, tracks citation paths, and stores PDFs in a local database. Helps agents review scientific evidence inside development workflows.
claude mcp add scholar-mcp -e S2_API_KEY=<S2_API_KEY> -e OPENALEX_API_KEY=<OPENALEX_API_KEY> -e CORE_API_KEY=<CORE_API_KEY> -e SCOPUS_API_KEY=<SCOPUS_API_KEY> -e DASHSCOPE_API_KEY=<DASHSCOPE_API_KEY> -e SCHOLAR_RERANK_API_KEY=<SCHOLAR_RERANK_API_KEY> -e EXA_API_KEY=<EXA_API_KEY> -e ZOTERO_API_KEY=<ZOTERO_API_KEY> -e NOTION_API_KEY=<NOTION_API_KEY> -- uvx scholar-mcpWhat Scholar MCP does
The server runs via stdio or Streamable HTTP, giving agents access to retrieval functions like search_papers and paper_info. Models query scientific indices using natural phrasing, filter results through rerankers, and inspect linked references. For direct examination, the read_paper tool parses document text, figures, and tables, defaulting to pages 1 through 10.
For broader literature mapping, the research profile adds build_paper_graph to evaluate PageRank centrality and draw citation networks using Mermaid syntax. Everything saves into an SQLite database with FTS5 search, enabling snapshot exports directly into Obsidian or Zotero vaults.
Running the packaged reranker requires uv with Python 3.12. Without provider credentials such as S2_API_KEY or OPENALEX_API_KEY, discovery relies on default semantic routes and encounters lower throughput limits.
What it helps with
- discover relevant scientific preprints using natural language queries
- map citation pathways with PageRank calculations and Mermaid graphs
- extract text and tables from specific publication pages without opening a browser
- store downloaded PDFs in a searchable local SQLite collection
- sync literature references and notes into Zotero or Obsidian
Example prompts for the agent
- Use search_papers to find recent preprints on reinforcement learning with human feedback.
- Map citation pathways for this paper with build_paper_graph and output a Mermaid diagram.
- Save this paper to the collection via download_paper and inspect the first five pages.
How to install Scholar MCP
Platform CLI
claude mcp add scholar-mcp -e S2_API_KEY=<S2_API_KEY> -e OPENALEX_API_KEY=<OPENALEX_API_KEY> -e CORE_API_KEY=<CORE_API_KEY> -e SCOPUS_API_KEY=<SCOPUS_API_KEY> -e DASHSCOPE_API_KEY=<DASHSCOPE_API_KEY> -e SCHOLAR_RERANK_API_KEY=<SCHOLAR_RERANK_API_KEY> -e EXA_API_KEY=<EXA_API_KEY> -e ZOTERO_API_KEY=<ZOTERO_API_KEY> -e NOTION_API_KEY=<NOTION_API_KEY> -- uvx scholar-mcpReplace <…> with your own values. 11 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from scholar-mcp <command>
Platform CLI
codex mcp add scholar-mcp --env SCHOLAR_DATA_DIR=<SCHOLAR_DATA_DIR> --env SCHOLAR_MCP_EXTENSIONS=<SCHOLAR_MCP_EXTENSIONS> --env S2_API_KEY=<S2_API_KEY> --env OPENALEX_API_KEY=<OPENALEX_API_KEY> --env CORE_API_KEY=<CORE_API_KEY> -- uvx scholar-mcpReplace <…> with your own values. 11 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from scholar-mcp <command>
Config file
{
"mcpServers": {
"scholar-mcp": {
"command": "uvx",
"args": [
"scholar-mcp"
],
"env": {
"SCHOLAR_DATA_DIR": "<SCHOLAR_DATA_DIR>",
"SCHOLAR_MCP_EXTENSIONS": "<SCHOLAR_MCP_EXTENSIONS>",
"S2_API_KEY": "<S2_API_KEY>",
"OPENALEX_API_KEY": "<OPENALEX_API_KEY>",
"CORE_API_KEY": "<CORE_API_KEY>"
}
}
}
}Replace <…> with your own values. 11 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from scholar-mcp <command>
Config file
{
"servers": {
"scholar-mcp": {
"type": "stdio",
"command": "uvx",
"args": [
"scholar-mcp"
],
"env": {
"SCHOLAR_DATA_DIR": "<SCHOLAR_DATA_DIR>",
"SCHOLAR_MCP_EXTENSIONS": "<SCHOLAR_MCP_EXTENSIONS>",
"S2_API_KEY": "<S2_API_KEY>",
"OPENALEX_API_KEY": "<OPENALEX_API_KEY>",
"CORE_API_KEY": "<CORE_API_KEY>"
}
}
}
}Replace <…> with your own values. 11 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from scholar-mcp <command>
Platform CLI
gemini mcp add -e SCHOLAR_DATA_DIR=<SCHOLAR_DATA_DIR> -e SCHOLAR_MCP_EXTENSIONS=<SCHOLAR_MCP_EXTENSIONS> -e S2_API_KEY=<S2_API_KEY> -e OPENALEX_API_KEY=<OPENALEX_API_KEY> -e CORE_API_KEY=<CORE_API_KEY> scholar-mcp uvx scholar-mcpReplace <…> with your own values. 11 more environment variables are in the server README. uvx runs the command named after the package; if the README names another one, use uvx --from scholar-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Scholar MCP
How do I install Scholar MCP in Claude Code?
Run this terminal command: `claude mcp add scholar-mcp -e S2_API_KEY=<S2_API_KEY> -e OPENALEX_API_KEY=<OPENALEX_API_KEY> -e CORE_API_KEY=<CORE_API_KEY> -e SCOPUS_API_KEY=<SCOPUS_API_KEY> -e DASHSCOPE_API_KEY=<DASHSCOPE_API_KEY> -e SCHOLAR_RERANK_API_KEY=<SCHOLAR_RERANK_API_KEY> -e EXA_API_KEY=<EXA_API_KEY> -e ZOTERO_API_KEY=<ZOTERO_API_KEY> -e NOTION_API_KEY=<NOTION_API_KEY> -- uvx scholar-mcp`.
What does Scholar MCP do?
It provides tools for literature retrieval, citation graph analysis, and PDF text extraction. The server indexes papers locally in an SQLite catalog and connects to external note-taking apps.
What dependencies does Scholar MCP require?
You need uv and Python 3.10 or newer, while the CPU reranker dependency specifically requires Python 3.12. Call the scholar://status resource to check active data providers and reranker health.
