Overview
These are Model Context Protocol (MCP) connectors that let AI assistants such as Claude and ChatGPT search or retrieve scholarly content directly. A content provider runs an MCP server, which MCP-compatible AI assistants can connect to.
1. AdisInsight in Claude (Springer Nature)
AdisInsight is Springer Nature’s database of pharmaceutical intelligence, covering medicines in development, clinical trials, safety reports, regulatory events and deals. Since 7 May 2026 institutions that subscribe to both Claude and AdisInsight can query it from inside Claude. The MCP server grounds Claude’s answers in licensed AdisInsight content and keeps provenance, access controls and usage tracking.
Notes & quotes
- “This integration reflects Springer Nature’s approach to giving its customers greater choice in how they access and work with our content.” Saskia Steinacker, Chief Product Officer, Springer Nature.
Sources
2. Article Galaxy MCP (Research Solutions)
Article Galaxy is Research Solutions’ service for acquiring research articles and managing the rights to use them. Research Solutions launched Article Galaxy MCP on 7 April 2026, so that researchers and AI agents can check access and reuse rights and order articles from inside their AI tools. Research Solutions says the MCP gives the user secure download links to licensed full text and does not deliver PDFs to the AI itself. Where using an article with AI needs AI and text-and-data-mining rights, these are bought separately, if they are available.
Notes & quotes
- “The Article Galaxy MCP now puts that entire system inside the AI environments researchers are already using to provide broader search capabilities and connectivity.” Roy Olivier, Chief Executive Officer, Research Solutions.
Sources
- Article Galaxy MCP, Research Solutions
- Research Solutions announces new Article Galaxy MCP, Research Solutions, 7 April 2026
3. bioRxiv and medRxiv connector (Anthropic)
Anthropic’s bioRxiv connector, listed in Claude’s connector directory since January 2026, lets Claude search, retrieve and summarise preprints from bioRxiv and medRxiv, the preprint servers for the life and health sciences run by openRxiv. Users can search by topic, funder or subject category, and check which preprints have since been published in journals. It needs no sign-in.
Sources
- bioRxiv connector, Claude
- Using the bioRxiv and medRxiv connector in Claude, Claude
- bioRxiv connector, PromptArmor
4. Cashmere
Cashmere licenses publishers’ content to AI applications for use in answering queries, with controls on how the content is used and reports on that use. Its publisher partners include Rockefeller University Press (from June 2026), Springer Publishing (April 2026) and Princeton University Press (August 2026), and it powers the KGL AI Gateway. Ian Mulvany describes Cashmere as the front runner among MCP services for publishers.
Notes & quotes
- Choice reports that Cashmere does not allow licensed content to be used to train large language models.
Sources
- Rockefeller University Press and Cashmere partner to bring authoritative life sciences research to AI applications, STM Publishing News, June 2026
- Springer Publishing Company and Cashmere announce partnership, Springer Publishing, 9 April 2026
- Announcing a partnership with Cashmere, Princeton University Press, 3 August 2026
- Academic Publishing Weekly, Choice, July 2026
- AI news, 7 August 2026, Ian Mulvany
5. Community-built servers for open scholarly databases
Developers have built many MCP servers for open scholarly databases such as OpenAlex, Semantic Scholar, arXiv, Crossref and Europe PMC, which researchers can install and connect to an AI assistant themselves. MCPBundles notes that arXiv does not offer an official MCP server, so these are community projects that vary in quality and maintenance.
Notes & quotes
- Aaron Tay notes that most of the guides to connecting AI tools to academic sources are for servers that connect to open content such as arXiv, Semantic Scholar and OpenAlex.
Sources
- MCP servers and academic search, Aaron Tay
- Academic research posts, MCPBundles
- OpenAlex MCP server, Glama
6. Consensus MCP
Consensus, an AI search tool for research, offers an MCP connector that lets users query its database of more than 220 million papers from inside AI assistants.
Sources
- Setting up the Consensus AI Connect (Consensus MCP Server), Consensus Help Center
- How to use Consensus MCP with AI agents for academic research, The Effortless Academic, 2 October 2026
7. Cortellis connectors (Clarivate)
Clarivate connected two of its Cortellis life sciences databases to Claude through MCP in 2026: Cortellis Regulatory Intelligence, announced on 10 March 2026, and Cortellis CMC Intelligence, which covers regulatory requirements for manufacturing medicines. Both give subscribers referenced answers from Clarivate’s curated regulatory content inside Claude, and subscribers can build agents that combine it with their own data.
Notes & quotes
- “By bringing Cortellis Regulatory Intelligence into Claude, we are expanding the reach of our trusted regulatory insights.” Henry Levy, President, Life Sciences & Healthcare, Clarivate.
Sources
- Clarivate expands access to trusted regulatory intelligence within Claude, Clarivate, 10 March 2026
- Cortellis CMC Intelligence connector, Claude
8. Dimensions and Altmetric MCP servers (Digital Science)
Digital Science launched two MCP servers for Dimensions, its database of linked research information, on 10 August 2026. Dimensions Semantic Search MCP searches life science publications, patents, clinical trials and drug labels by concept, and Dimensions Analytics MCP gives AI agents access to more than 430 million linked records. Research Information reports that existing Dimensions API customers can use them without an additional licence.
Digital Science launched the Altmetric MCP on 6 August 2026. It connects AI agents to Altmetric’s data on attention to research in news media, social media, policy documents, clinical guidelines, patents and blogs. An agent can use it to retrieve a publication’s Altmetric attention score, engagement breakdown and sentiment, to find which news outlets, policy bodies and online platforms are discussing research on a topic, and to aggregate engagement across a portfolio of publications for reporting. Digital Science says it works with Claude, ChatGPT, VS Code, Cursor and other AI platforms, and that it is meant for medical affairs teams, researchers, institutions, funders and publishers. In October 2026 Digital Science described the Altmetric data as more than 256 million mentions of more than 24 million research outputs.
Notes & quotes
- “What they need is an authoritative bridge between AI and research intelligence.” Sebastian Schmidt, EVP Enterprise, Digital Science.
- “Dimensions Semantic Search MCP is designed specifically for the complexity of life science terminology.” Peter Haase, VP Knowledge Graph Technologies, Digital Science.
- “This information is all in a single place and is often available just hours post publication.” Carlos Areia, Senior Data Scientist, Digital Science.
Sources
- Digital Science connects AI agents to world-leading research data with new Dimensions MCP servers, Digital Science, 10 August 2026
- Digital Science launches Dimensions MCP servers, Research Information, August 2026
- Digital Science brings real-world research impact data into AI workflows with new Altmetric MCP, Digital Science, 6 August 2026
- The Download, Digital Science, LinkedIn, 23 October 2026
9. Elicit API and MCP server
In July 2026 Elicit, an AI research assistant, made its API and MCP server generally available. The search API covers 138 million academic papers and 545,000 clinical trials, and the reports API returns a cited report on a research question. A new systematic review API runs a full systematic review, with the user setting the search strategy, screening criteria, data extraction and reporting.
Sources
10. Elsevier connectors
Elsevier offers connectors for corporate R&D teams as add-ons to licences for its datasets, so that they can query Elsevier data through MCP or APIs in their AI tools. There are connectors for full-text journal articles in ScienceDirect, chemistry data in Reaxys, and drug safety and pharmacokinetic data in PharmaPendium. Elsevier says the connectors match compounds and terms across synonyms and registry numbers, and that results link back to source records.
Benefits & outcomes
- Elsevier reports that a global pharmaceutical company using connectors achieved 30% better retrieval, 25% lower token cost and 20% better insight quality.
Sources
11. GetFTR access checks
GetFTR, which connects users to the full text of academic content they have access to, runs an MCP server that AI research tools can ask whether a user has access to a particular item. The tool can then send the user to the version of record they are entitled to read.
Sources
12. KGL AI Gateway (KGL and Cashmere)
The KGL AI Gateway, from KnowledgeWorks Global (KGL) and powered by Cashmere’s licence-aware AI infrastructure, connects publishers’ content to approved AI applications while the publisher keeps control of rights, attribution and payment. KGL and Cashmere announced their partnership in October 2025.
Sources
- KGL AI Gateway, KGL
- Cashmere and KGL partner to power the commercial and technical rails for premium content in AI, KGL, 15 October 2025
13. Meridian (Cashmere, with Cambridge, Johns Hopkins and Princeton University Presses)
Meridian is an MCP gateway that gives AI assistants authorised access to library-licensed scholarly content through a single integration. Cambridge University Press, Johns Hopkins University Press and Princeton University Press announced on 4 November 2026 that they are its founding publishers.
Cashmere convened the initiative, runs it and provides the platform. The founders say that each answer an AI assistant draws from Meridian is grounded in peer-reviewed content, cites it and links to the publisher’s version of record through the reader’s library access.
Libraries receive COUNTER-aligned reports on how their licensed collections are used in AI-assisted research, and publishers receive statistics on the questions their catalogues answered and how often they are cited across AI tools. Content reached through Meridian may not be used to train or fine-tune AI models, or be sublicensed or redistributed. Publishers take part by choice, on a non-exclusive basis, and can withdraw titles at any time.
The founding presses and Cashmere form an Advisory Council, which becomes the board of directors when Meridian incorporates as an independent not-for-profit. The founding publishers are adding their catalogues in November 2026, a closed beta for selected institutions and libraries is planned for later in autumn 2026, and Cashmere has committed its platform for an initial 24-month evaluation.
Notes & quotes
- Presenting Meridian in November 2026, Jonathan Woahn of Cashmere said that researchers are unsure whether uploading licensed books to AI systems is permitted, and that usage statistics cannot tell a cursory download from hours of substantive engagement.
- “Meridian is shared infrastructure that belongs to the scholarly community.” Jonathan Munk, CEO, Cashmere.
- Research Information reports that Meridian won the award for best design at the Charleston Premiers.
Sources
- Cambridge University Press, Johns Hopkins University Press, Princeton University Press and Cashmere to launch Meridian, STM Publishing News, 4 November 2026
- Charleston Premiers: 10 approaches to the changing research landscape, Research Information, 5 November 2026
14. Nexus Connect (Clarivate)
Clarivate announced Nexus Connect on 28 April 2026. It is a single university-branded connector that brings Clarivate content, the library’s licensed resources and services from other academic vendors into AI chat agents, applying the institution’s subscriptions. Libraries choose which content and services it shows. Clarivate said five universities in North America and Asia-Pacific would be among the first to use it, with an early access programme planned for July 2026.
Notes & quotes
- “User expectations are changing fast as they want the speed and ease of AI as part of how they study and research.” Shirley Wong, University Librarian, The Hong Kong Polytechnic University, quoted in Clarivate’s announcement.
Sources
- Clarivate introduces Nexus Connect, the first institutional AI gateway to trusted research and learning, Clarivate, 28 April 2026
- Nexus Connect, Clarivate
15. Nexus MCP (Annexis)
Nexus MCP, from Annexis, is an integration service for AI teams that connects agents to scholarly records and keeps evidence of how the records are accessed and used. Annexis says it “measures agentic reuse with provenance and evidence”, and that it cannot add missing author, institution, funder or reference information to a record.
Sources
16. Paperclip (GXL)
Paperclip, from GXL, gives AI agents access to full-text papers, clinical trials, regulatory documents, patents and biological databases through an MCP endpoint.
Sources
17. Publication Tracker MCP server (Appetence)
Appetence launched an MCP server for Publication Tracker, its open-source (meta)data repository for publishers, in the week of 1 October 2026. It lets a publisher connect an AI assistant such as Claude to the ingested, extracted and full-text (meta)data held in Publication Tracker, so that the assistant can look up answers, run analyses or act as an agent working directly on the data.
Sources
18. PubMed connector (Anthropic)
Anthropic’s PubMed connector, part of Claude for Life Sciences since October 2025, lets Claude search PubMed and read full-text articles in PubMed Central from inside a conversation. It connects to the public database, so users do not need to sign in.
Sources
- Using the PubMed connector in Claude, Claude
- Anthropic launches Claude for Life Sciences with Benchling, PubMed integration, Maginative
19. Scite MCP and publisher gateway (Research Solutions)
Scite’s MCP connectors, including a Claude connector launched on 29 April 2026, let researchers search Scite’s index of papers and citations, and read full-text papers, from inside AI assistants, with Article Galaxy supplying the rights for AI use. Scite also offers publishers a gateway into this channel, which its co-founder Josh Nicholson says has no indexing cost for publishers.
MDPI and Research Solutions announced on 6 October 2026 that MDPI’s journal portfolio, of more than 500 open access journals, is indexed in Scite and delivered through the Scite MCP, so that researchers can find MDPI articles in Claude, ChatGPT, Microsoft Copilot and other MCP-compatible AI tools. MDPI articles appear in AI responses with citation context from Scite’s Smart Citations and links to the version of record on MDPI’s platform. MDPI receives article-level reports showing which articles are accessed, by which organisations and through which AI platforms. Research Solutions handles the indexing, transformation, authentication and integration.
Benefits & outcomes
- Josh Nicholson reports that AI assistants read 8.1 million research papers through Scite in September 2026, and that nearly half hit a paywall.
- Research Solutions reports that the Scite MCP has served more than 22 million requests since its launch earlier in 2026, with monthly volume growing by more than 30% month on month.
Notes & quotes
- Aaron Tay notes that Scite’s MCP server can search for papers with many contrasting citations, which Consensus cannot.
- A scientist at a large manufacturer who uses AI agents with Scite to check code against the literature says it has “cut down these steps and increased my throughput”, according to Nicholson.
- Dr Constanze Schelhorn, Chief Publishing Officer at MDPI, says “it is important that AI-generated answers are grounded in the scholarly record”.
Sources
- Josh Nicholson on AI reads through Scite, LinkedIn, 15 October 2026
- Josh Nicholson on Scite’s publisher gateway, LinkedIn, 12 October 2026
- What if Claude or ChatGPT could search academic databases for you, and then do something with the results?, Aaron Tay, Singapore Management University Libraries, 2026
- Research Solutions’ Scite & MDPI collaborate to make open access research discoverable, attributed, & measurable inside AI tools, PR Newswire, 6 October 2026
20. Silverchair Discovery Bridge
Silverchair launched Discovery Bridge on 10 February 2026. It connects AI assistants to one publisher’s full-text content on Silverchair’s publishing platform, using the publisher’s existing subscription rules. Retracted articles are left out of results. Silverchair says it also lets publishers sell corporate content bundles and pay-per-use licences. In October 2026 Silverchair added Research Bridge, which puts several publishers’ content behind one shared connector.
Notes & quotes
- “Our Discovery Bridge gives publishers the necessary infrastructure to deliver their insights and content into AI solutions while protecting their valuable intellectual property and creating new revenue streams.” Will Schweitzer, CEO, Silverchair.
Sources
- Silverchair launches the Discovery Bridge MCP to connect scholarly content with AI-powered research workflows, Silverchair, 10 February 2026
- The Silverchair Lab, Silverchair
21. Tessera (Molecular Connections)
Molecular Connections launched Tessera in August 2026 as a platform that lets publishers govern, understand and license their content for AI applications, built on its work in semantic enrichment and knowledge graphs. Ian Mulvany describes it as an addition to the MCP space for publishers.
Sources
- Molecular Connections introduces Tessera, Molecular Connections, 3 August 2026
- AI news, 7 August 2026, Ian Mulvany
22. Wiley Scholar Gateway
Wiley’s Scholar Gateway, launched in October 2025 as the Wiley AI Gateway, connects AI assistants to Wiley journals, so that answers draw on peer-reviewed sources with DOI links. Wiley says it covers Wiley content only for now, with other publishers’ content to be added, and users have to sign in with Wiley.
Sources
23. Yale Library AI Connector
Yale Library’s AI Connector connects Claude and ChatGPT to the library’s catalogue and digital collections, covering almost 15 million catalogue records and nearly 185,000 digital collection records. Yale opened it to the Yale community for testing on 15 October 2026. Licensed articles and databases such as JSTOR and EBSCO are not included.
Notes & quotes
- “Students at Yale already use AI chatbots for their academic work.” Chelsea Fitzgerald, Yale Library.
- “We are ahead of other institutions in implementing this type of tool.” Michael Appleby, Director of Software Engineering, Library IT, Yale Library.
- Ten students tested the connector for a week in autumn 2025. They asked for sample prompts and for help telling library sources apart from open web sources, and found that users have to ask the chatbot to use Yale Library resources. The library wrote quick-start guides and sample prompts in response.
Sources
This entry was written with the help of AI, from the sources listed. Please check the sources before relying on it, and let us know if anything needs correcting.
