START HERE — WHAT IS MCP?
1. Demystifying Model Context Protocol (MCP) (Silverchair · Dec 2025)We may be biased, but we think this is the best first read for publishers. Explains MCP through the familiar lens of APIs, covers how it differs, and frames the strategic questions publishers need to be thinking about — access, attribution, licensing, metadata standards — without getting lost in the technical weeds.
2. What is Model Context Protocol (MCP)? A guide (Google Cloud · 2025)
Google’s accessible overview walks through the host-client-server model with concrete examples, and draws useful distinctions between MCP and RAG — a comparison that matters a lot for how publishers think about their content infrastructure.
3. Model Context Protocol (Wikipedia · Updated Apr 2026)
A useful reference for the origin story and ecosystem snapshot: who created it, who adopted it (OpenAI, Google DeepMind, Microsoft), and how MCP moved from an Anthropic initiative to an industry standard under the Linux Foundation’s Agentic AI Foundation.
MCP IN SCHOLARLY PUBLISHING CONTEXT
4. AI readiness and the new value equation in scholarly publishing (The Scholarly Kitchen · Feb 2026)Examines what it actually means for content to be “AI-ready” in an MCP world. Makes a strong case that raw XML and PDFs are insufficient — publishers need to think seriously about chunking, semantic tagging, and ontologies before exposing content through MCP-enabled integrations.
5. The User Has Changed. Has Scholarly Publishing? (The Scholarly Kitchen · May 2026) A comprehensive treatment of the MCP moment based on the Spring 2026 Platform Strategies Webinar Series: the JAMA bot-traffic case study, OUP's engagement data (30% longer sessions, 40% click-through on conversational follow-up), Jeremy Little's framing of where the infrastructure and provider gaps actually are, and Tasha Mellins-Cohen's COUNTER warning about negotiating before the norms calcify. The stat that 47% of a publisher audience had heard of MCP but couldn't explain it, and 27% hadn't encountered the term, gives this real currency as a "here's where the industry actually is" data point.
6. Model Context Protocol (MCP): landscape, security threats, and future research directions (ACM Transactions on Software Engineering and Methodology · Feb 2026)
The most rigorous technical treatment of MCP on this list — a peer-reviewed survey that maps the full lifecycle of an MCP server across four phases (creation, deployment, operation, and maintenance), and systematically examines the security vulnerabilities that come with it, including prompt injection, over-permissioned tool access, and authentication gaps. Not a light read, but publishers evaluating MCP investments will find the security section particularly relevant: it's the most thorough documentation available of the risks that need to be designed around, and it provides a useful counterweight to the more optimistic industry coverage.
7. AI isn’t going to pay for content — at least not how you’re hoping it will (The Scholarly Kitchen · Jan 2026)
Part one of this two-part series argues that AI training licensing is a false revenue hope — one-time transactions with depreciating assets — and that the real opportunity lies in inference: content accessed at the moment a user asks a question. Part 2 offers the business case for inference monetization — content accessed through APIs and MCP at the moment AI applies it — is the durable opportunity.
8. There’s an elephant in the room, but not in your usage reports (The Scholarly Kitchen · Feb 2026)
Essential reading. Argues that publisher usage hasn’t disappeared — it’s been displaced to agentic platforms that current COUNTER metrics can’t see. Uses real examples from to show how MCP is at the center of this shift and what publishers need to do about it.
9. Usage data in an AI world: key shifts, challenges, and JSTOR’s response (JSTOR · Jan 2026)
Continuing the usage theme, a grounded look at how AI-mediated research is breaking traditional usage metrics. Covers COUNTER’s emerging work on AI-specific reporting and explains why the “cost-per-use” model breaks down when an AI agent reads an article on a researcher’s behalf.
10. Key takeaways from COUNTER's AI usage metrics consultation (COUNTER Metrics · Mar 2026)
After a timely community consultation that drew near-universal support, COUNTER's AI working group has committed to adding Access_Method: Agent to distinguish machine usage from human usage across Platform, Database, Title, and Item Reports — and is now developing new AI-specific metric types. For publishers managing library relationships and demonstrating institutional value, this is where the abstract conversation about MCP and agentic access becomes a reporting reality.
11. Scholarly AI Search Shortcomings and the Need for Better Metadata (The Scholarly Kitchen · May 2026) Peter Webster (librarian) makes the case that AI scholarly search is failing not just because of hallucination but because of incomplete and inconsistent metadata — and that knowledge graphs and vector embeddings derived from full text are the path forward. MCP comes in as one of the publisher-side frameworks being used to connect AI searches to content in a controlled way.
INDUSTRY SIGNALS & STRATEGIC QUESTIONS
12. A beginner’s guide to MCP: what it is and why it matters (Cloudera · Jul 2025)Zooming back out, this article takes the foundational concepts and applies them to enterprise strategy — covering how MCP keeps AI on task, enforces access boundaries, and operates across distributed environments. The enterprise framing translates well to publisher concerns around content governance and controlled AI access.
13. Ask the right questions: A technologist's advice on MCPs (Silverchair, May 2026)
Silverchair’s Scott Bouchard observes that as publishers explore MCPs, much of the conversation has focused on opportunity; not enough has focused on mechanics. That gap matters because MCP is not just a new distribution channel. It is a layer of infrastructure with direct bearing on content authority, access rights, competitive positioning, and trust. Before the industry commits to building, buying, or partnering around it, there are questions worth sitting with.
14. The 2026-07-28 MCP Specification Release Candidate (MCP Official Blog · May 2026) This represents the largest protocol revision since launch. The headline change is MCP going fully stateless, which has major infrastructure implications: no more sticky sessions or shared session stores, runs on ordinary HTTP load balancers. Also introduces the Extensions framework, Tasks, MCP Apps (server-rendered UIs), and authorization hardening. The final spec publishes July 28. For publishers evaluating MCP timing and architecture, this is essential context.
15. A Platform Comes Apart: Part 1(The Scholarly Kitchen · Jun 2026) Based on a PurePub.ai panel with Adam Hyde (Purepub.ai), Stuart Leitch (Silverchair), Hong Zhou (KnowledgeWorks), and Paul Shannon (eLife), it argues that platforms aren't dying — they're splitting into a trusted headless core and a contested agentic layer above it. Stuart's framing is notable: platforms broken into "primitives," the agentic layer "up for grabs," the seat model giving way to consumption pricing, and his instinct to "disrupt ourselves at that same speed" before someone else does. The strategic questions raised — where value sits as agents increasingly mediate discovery, review, and production — are exactly what publishers need to be thinking through. Though this piece doesn't mention MCP by name, it's the most direct treatment of the structural question MCP is forcing on publishers.
Wondering what this looks like in practice? Visit Silverchair's Discovery Bridge MCP page for details on the key benefits of MCP and scholarly publishing-specific uses case, including corporate content bundles and more. Want to keep learning? Join industry leaders in DC this Fall for Platform Strategies 2026, where we'll tackle the biggest topics in scholarly technology.