OmniScholar — Claude Code MCP server: research and setup
An MCP server for literature search, Zotero reading, PDF analysis, citations, and research materials. Use it when an agent needs a research workflow.
claude mcp add omnischolar -- uvx luffysolution-omnischolarWhat OmniScholar does
OmniScholar gives an agent 39 tools for research work. It can search Semantic Scholar, OpenAlex, PubMed/PMC, arXiv, Crossref, Unpaywall, easyScholar, Google Scholar, and Google Patents. It also retrieves paper details, author data, references, citing papers, datasets, and journal metrics. Zotero content is read without changing the library, while Materials Project records can be exported as JSON, CSV, Markdown, or CIF.
After your approval, a selected PDF is sent to MinerU to extract prose, formulas, tables, and images. The server can use that material for focused or full analysis, save notes to a regular folder or an Obsidian Vault, check citation candidates, and call a configured image service for scientific diagrams. Python 3.11 or newer is required to run the server.
The MCP process runs locally through uvx. Service credentials go into omnischolar.config.json or an environment variable, and services that need them require an API key. PDF uploads to MinerU require confirmation first; Zotero remains read-only.
What it helps with
- search scientific papers across several indexes
- combine Zotero notes and annotations without editing the library
- extract formulas, tables, images, and text from a PDF
- verify citation candidates and format references
- export Materials Project records to CSV or CIF
Example prompts for the agent
- Find 5 recent reviews about solid-state battery interfaces, deduplicate them by DOI, and list open-access versions.
- Search my Zotero library for this DOI, summarize its notes and annotations, and leave the library unchanged.
- After I approve the upload, process this PDF with MinerU and save the reading notes to my Obsidian Vault.
How to install OmniScholar
Platform CLI
claude mcp add omnischolar -- uvx luffysolution-omnischolaruvx runs the command named after the package; if the README names another one, use uvx --from luffysolution-omnischolar <command>
Platform CLI
codex mcp add omnischolar -- uvx luffysolution-omnischolaruvx runs the command named after the package; if the README names another one, use uvx --from luffysolution-omnischolar <command>
Config file
{
"mcpServers": {
"omnischolar": {
"command": "uvx",
"args": [
"luffysolution-omnischolar"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from luffysolution-omnischolar <command>
Config file
{
"servers": {
"omnischolar": {
"type": "stdio",
"command": "uvx",
"args": [
"luffysolution-omnischolar"
]
}
}
}uvx runs the command named after the package; if the README names another one, use uvx --from luffysolution-omnischolar <command>
Platform CLI
gemini mcp add omnischolar uvx luffysolution-omnischolaruvx runs the command named after the package; if the README names another one, use uvx --from luffysolution-omnischolar <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 OmniScholar
How do I install OmniScholar in Claude Code?
Run claude mcp add omnischolar -- uvx luffysolution-omnischolar.
What does OmniScholar do?
It searches academic sources, reads Zotero data, processes approved PDFs with MinerU, checks citation details, and works with Materials Project records. Outputs can be saved in Markdown and other supported formats.
What does OmniScholar require to run?
Python 3.11 or newer is required. Connected services may also need an API key supplied through the configuration or an environment variable.
