Research MCP

# 🧠 Research MCP - Model Context Protocol Research Assistant [](https://www.npmjs.com/package/researchmcp) [](https://github.com/gyash1512/ResearchMCP) [](https://opensource.org/licenses/MIT) A complete Model Context Protocol (MCP)-based Research Assistant that enables LLMs to fetch, analyze, and summarize academic research papers in real-time from multiple trusted sources: **arXiv**, **Semantic Scholar**, and **PubMed**. ## 🔍 Overview The Research MCP system provides standardized access to academic research databases through three specialized MCP servers. Each server implements the MCP specification, allowing AI assistants to query live research data, process results, and return structured insights like summaries, comparisons, and citations. ## ✨ Features - 📚 **Multi-Source Search**: Query arXiv, Semantic Scholar, and PubMed simultaneously - 🔄 **Automatic Deduplication**: Smart paper matching across different sources - 📊 **Citation Analysis**: Track citation counts and influential papers - 📝 **BibTeX Generation**: Automatic citation formatting for all sources - ⚡ **Rate Limiting**: Built-in request throttling to respect API limits - 🎯 **Advanced Filtering**: Filter by year, author, venue, and more - 🔍 **Full Metadata**: Complete paper information including abstracts, authors, and links ## 🏗️ Architecture ``` ┌─────────────────────────────────────────────────────────────┐ │ LLM Client │ │ (Issues natural language queries) │ └────────────────┬────────────────────────────────────────────┘ │ │ MCP Protocol │ ┌────────────────┴───────────────────────────────────────────┐ │ MCP Servers │ │ ┌──────────┐ ┌──────────────┐ ┌──────────┐ │ │ │ arXiv │ │ Semantic │ │ PubMed │ │ │ │ Server │ │ Scholar │ │ Server │ │ │ └────┬─────┘ └──────┬───────┘ └────┬─────┘ │ └───────┼────────────────┼───────────────┼───────────────────┘ │ │ │ │ │ │ ┌───────┴────────────────┴───────────────┴───────────────────┐ │ External APIs │ │ arXiv API Semantic Scholar API PubMed E-utils │ └────────────────────────────────────────────────────────────┘ ``` ## 📋 Prerequisites - **Node.js** 18+ - An MCP-compatible client (Claude Desktop, Cline, etc.) ## 🚀 Installation ### Quick Start with npx (Recommended) **No installation or API keys needed!** Just add to your MCP client configuration: ```json { "mcpServers": { "research-arxiv": { "command": "npx", "args": ["-y", "researchmcp", "arxiv"] }, "research-semantic-scholar": { "command": "npx", "args": ["-y", "researchmcp", "semantic"] }, "research-pubmed": { "command": "npx", "args": ["-y", "researchmcp", "pubmed"] } } } ``` **That's it!** All three servers work perfectly without any API keys or configuration. --- ### Local Development For contributing or modifying the code: ```bash git clone https://github.com/gyash1512/ResearchMCP.git cd ResearchMCP npm install npm run build ``` ## 🎮 Usage ### Using with npx (Recommended) Just configure in your MCP client - that's it! No API keys needed. ### Local Development Start servers individually for testing: ```bash npm run start:arxiv npm run start:semantic npm run start:pubmed ``` ### MCP Configuration **Simple setup - no API keys required:** ```json { "mcpServers": { "research-arxiv": { "command": "npx", "args": ["-y", "researchmcp", "arxiv"] }, "research-semantic-scholar": { "command": "npx", "args": ["-y", "researchmcp", "semantic"] }, "research-pubmed": { "command": "npx", "args": ["-y", "researchmcp", "pubmed"] } } } ``` <details> <summary><b>Optional:</b> Add API keys for higher rate limits (only if needed)</summary> ```json { "mcpServers": { "research-semantic-scholar": { "command": "npx", "args": ["-y", "researchmcp", "semantic"], "env": { "SEMANTIC_SCHOLAR_API_KEY": "your_key_here" } }, "research-pubmed": { "command": "npx", "args": ["-y", "researchmcp", "pubmed"], "env": { "PUBMED_API_KEY": "your_key_here", "PUBMED_EMAIL": "your_email@example.com" } } } } ``` </details> <details> <summary>For local development</summary> ```json { "mcpServers": { "research-arxiv": { "command": "node", "args": ["./dist/servers/arxiv-server.js"], "cwd": "/absolute/path/to/ResearchMCP" }, "research-semantic-scholar": { "command": "node", "args": ["./dist/servers/semantic-scholar-server.js"], "cwd": "/absolute/path/to/ResearchMCP" }, "research-pubmed": { "command": "node", "args": ["./dist/servers/pubmed-server.js"], "cwd": "/absolute/path/to/ResearchMCP" } } } ``` > **Note**: Replace `/absolute/path/to/ResearchMCP` with your actual project path. </details> ## 📚 Available Tools ### arXiv Server #### `search_arxiv` Search for papers on arXiv by keyword, author, or subject. **Parameters:** - `query` (string, required): Search query - `maxResults` (number, optional): Max results (default: 10, max: 100) - `startYear` (number, optional): Filter by start year - `endYear` (number, optional): Filter by end year - `author` (string, optional): Filter by author name - `sortBy` (string, optional): Sort by relevance, lastUpdatedDate, or submittedDate **Example:** ```json { "query": "quantum computing", "maxResults": 5, "startYear": 2023, "sortBy": "relevance" } ``` #### `get_arxiv_paper` Get detailed information about a specific arXiv paper by ID. **Parameters:** - `arxivId` (string, required): arXiv paper ID (e.g., "2301.12345") #### `arxiv_to_bibtex` Convert arXiv paper to BibTeX format. **Parameters:** - `arxivId` (string, required): arXiv paper ID --- ### Semantic Scholar Server #### `search_semantic_scholar` Search for papers with citation information. **Parameters:** - `query` (string, required): Search query - `maxResults` (number, optional): Max results (default: 10, max: 100) - `startYear` (number, optional): Filter by start year - `endYear` (number, optional): Filter by end year **Example:** ```json { "query": "transformer architecture", "maxResults": 10, "startYear": 2023 } ``` #### `get_semantic_scholar_paper` Get paper by Semantic Scholar ID or DOI. **Parameters:** - `identifier` (string, required): Paper ID or DOI #### `get_paper_citations` Get papers that cite a specific paper. **Parameters:** - `paperId` (string, required): Semantic Scholar paper ID - `maxResults` (number, optional): Max citing papers (default: 10, max: 100) #### `semantic_scholar_to_bibtex` Convert paper to BibTeX format. **Parameters:** - `identifier` (string, required): Paper ID or DOI --- ### PubMed Server #### `search_pubmed` Search biomedical and life sciences papers. **Parameters:** - `query` (string, required): Search query (supports MeSH terms) - `maxResults` (number, optional): Max results (default: 10, max: 100) - `startYear` (number, optional): Filter by start year - `endYear` (number, optional): Filter by end year **Example:** ```json { "query": "cancer treatment", "maxResults": 5, "startYear": 2022 } ``` #### `get_pubmed_paper` Get paper by PMID. **Parameters:** - `pmid` (string, required): PubMed ID #### `pubmed_to_bibtex` Convert paper to BibTeX format. **Parameters:** - `pmid` (string, required): PubMed ID ## 💡 Example Queries ### Example 1: Multi-Source Research Query **Query:** "Find recent papers on federated learning in healthcare" **Workflow:** 1. Search arXiv: `search_arxiv` with query "federated learning healthcare", startYear: 2023 2. Search Semantic Scholar: `search_semantic_scholar` with same parameters 3. Search PubMed: `search_pubmed` with same parameters 4. Combine and deduplicate results 5. Sort by citation count and relevance 6. Generate summary with top 5 papers **Expected Output:** - Comprehensive list of papers from all sources - Deduplicated results - Citation counts where available - Links to full papers - BibTeX citations ### Example 2: Most Cited Paper **Query:** "What's the most cited 2023 paper on quantum machine learning?" **Workflow:** 1. Call `search_semantic_scholar`: ```json { "query": "quantum machine learning", "maxResults": 50, "startYear": 2023, "endYear": 2023 } ``` 2. Sort results by `citationCount` 3. Get detailed info with `get_semantic_scholar_paper` 4. Generate BibTeX with `semantic_scholar_to_bibtex` **Expected Output:** - Paper title and authors - Citation count and venue - Abstract and key findings - BibTeX citation - Link to paper ### Example 3: Research Trend Analysis **Query:** "Summarize transformer innovations after 2023" **Workflow:** 1. Search multiple sources for "transformer architecture" papers after 2023 2. Extract key information from abstracts 3. Identify common themes and methods 4. Generate trend analysis 5. Provide top papers with citations **Expected Output:** - Overview of key innovations - Timeline of developments - Most influential papers - Citation network analysis - Recommended reading list ### Example 4: Citation Network **Query:** "Find papers citing 'Attention is All You Need'" **Workflow:** 1. Find original paper: `search_semantic_scholar` with title 2. Get paper ID from results 3. Call `get_paper_citations` with the paper ID 4. Filter by year/relevance 5. Generate summary of citing papers **Expected Output:** - List of papers that cite the original work - Citation contexts - Related research directions - Impact analysis ## 🔧 API Response Schemas ### arXiv Paper Object ```typescript { id: string; // arXiv ID (e.g., "2301.12345") title: string; authors: string[]; abstract: string; published: string; // ISO date updated: string; // ISO date url: string; // Paper URL pdfUrl: string; // PDF download URL categories: string[]; // Subject categories primaryCategory: string; } ``` ### Semantic Scholar Paper Object ```typescript { paperId: string; title: string; abstract: string | null; year: number | null; authors: Array<{ authorId: string; name: string; }>; citationCount: number; referenceCount: number; influentialCitationCount: number; url: string; venue: string | null; publicationDate: string | null; } ``` ### PubMed Paper Object ```typescript { pmid: string; // PubMed ID title: string; abstract: string; authors: string[]; journal: string; year: string; doi: string | null; url: string; publicationTypes: string[]; meshTerms: string[]; // Medical Subject Headings } ``` ## 🛡️ Rate Limiting All servers work great without API keys: | Server | Default Rate | With API Key | Do You Need Keys? | |--------|--------------|--------------|-------------------| | **arXiv** | 3 req/sec | N/A | ❌ No - works perfectly! | | **Semantic Scholar** | 1-3 req/sec | 10 req/sec | ❌ No - unless making 100+ queries/min | | **PubMed** | 3 req/sec | 10 req/sec | ❌ No - unless making 100+ queries/min | **Recommendation:** Start without any API keys. Only add them if you hit rate limits. ## 🔒 Security Notes - **No API keys needed** - all servers work out of the box - If using API keys, pass via MCP config `env` section (see optional config above) - Never commit API keys to version control - Respect API rate limits and terms of service ## 📖 MCP Specification Compliance This implementation follows the [Model Context Protocol specification](https://modelcontextprotocol.io/): - ✅ Standard tool definition schema - ✅ JSON-based request/response format - ✅ Error handling with proper status codes - ✅ Resource management and cleanup - ✅ Stdio transport for client communication ## 🤝 Contributing Contributions are welcome! Areas for improvement: - Additional research sources (IEEE, ACM, etc.) - Advanced filtering and ranking algorithms - Paper recommendation system - Citation graph visualization - Full-text analysis capabilities ## 📄 License MIT License - See LICENSE file for details ## 🙏 Acknowledgments - **arXiv** for open access to research papers - **Semantic Scholar** for citation data and API - **PubMed/NCBI** for biomedical research database - **Model Context Protocol** team for the MCP specification ## 📞 Support For issues, questions, or contributions: - Open an issue on GitHub - Check API documentation for each service - Review MCP specification for protocol details --- **Built with ❤️ using TypeScript and the Model Context Protocol**