We’ve reviewed 6 categories of Science & Research MCP servers, evaluating over 170 individual servers across academic literature search, scientific computing, bioinformatics, chemistry, astronomy, and geospatial mapping. Each review covers architecture patterns, star counts, tool inventories, known issues, and honest ratings.

Science & Research is where MCP meets the lab bench. The pattern here differs from enterprise categories — instead of vendor-backed official servers, most science MCP servers come from individual researchers and academic groups. The exceptions matter: MathWorks ships an official MATLAB server, Mapbox ships 2 official geospatial servers, and Anthropic maintains a life-sciences marketplace. Community projects like mcp.science (12+ bundled servers) and the Augmented-Nature organization (10+ biology database servers) show that researchers are building their own AI infrastructure.

Academic & General Research

The broadest science category covers paper search, citation analysis, and general research workflows. arXiv MCP (2,400 stars) is the most popular science-specific MCP server by a wide margin. Multi-source aggregators like paper-search-mcp (7 academic sources) and mcp.science (12+ servers) provide comprehensive coverage. Astronomy fills a fascinating niche — NASA-MCP-server covers 20+ data sources, while astro_mcp provides access to 40+ astronomical surveys.

Review Rating Key Servers
Science & Research 3.5/5 arXiv MCP (2,400 stars, paper search/download/analysis), paper-search-mcp (796 stars, 7 sources), mcp.science (117 stars, 12+ servers), UniProt MCP (26 tools) — 35+ servers
Astronomy & Space Science 3.5/5 NASA-MCP-server (81 stars, 20+ APIs), astro_mcp (40+ surveys via astroquery), CelestialMCP (117k stars catalog), satellite tracking — 12+ servers

Scientific Computing & Mathematics

The computational backbone. MATLAB’s official MCP server (178 stars) is one of the few major scientific computing vendors to support MCP. R statistics is surprisingly mature — RMCP provides 52 tools across 429 packages. Julia has three MCP implementations. Wolfram and SymPy handle symbolic math. Globus Labs connects AI agents to supercomputers at ALCF and NERSC. OpenFOAM provides CFD simulation.

Review Rating Key Servers
Scientific Computing & Mathematics 3.5/5 MATLAB official (178 stars, MathWorks), RMCP (197 stars, 52 R tools, 429 packages), mcp.science (117 stars, DFT/Materials Project), SymPy MCP (41 stars), Globus HPC — 30+ servers

Life Sciences & Chemistry

The deepest science MCP ecosystem. Bioinformatics leads with 40+ servers — genomoncology/biomcp (241 stars) and Anthropic’s life-sciences marketplace (259 stars) provide integrated biomedical querying. The Augmented-Nature organization maintains 10+ servers wrapping every major biological database. Chemistry covers the full pipeline from cheminformatics (RDKit) through molecular docking (AutoDock Vina) to molecular dynamics (LAMMPS). Drug discovery has end-to-end coverage from target identification through patent search.

Review Rating Key Servers
Bioinformatics & Life Sciences 4/5 biomcp (241 stars, biomedical queries), Anthropic life-sciences (259 stars), PubMed MCP (105 stars), ChEMBL MCP (78 stars, 22 drug discovery tools) — 40+ servers
Chemistry & Molecular Modeling 3.5/5 ChatMol molecule-mcp (~89 stars, PyMOL/ChimeraX), RDKit MCP (cheminformatics), PubChem (110M+ compounds), DrugBank (17,430+ drugs), Vina docking — 25+ servers

Geospatial & Mapping

The highest-rated science subcategory. GIS platforms, satellite imagery, and mapping tools with official vendor support from Mapbox and Google Earth Engine. Covers the full geospatial stack from OpenStreetMap to NASA Earthdata.

Review Rating Key Servers
Geospatial & Mapping 4.5/5 Mapbox (2 official servers), Google Earth Engine, NASA Earthdata, QGIS, OpenStreetMap, GIS libraries — 30+ servers