Loading…
The next generation of MCP
CloudflareMatt Carey
Summary
Model Context Protocol previously required stateful connections between clients and remote servers, creating operational complexity around sticky sessions, message replay, and open streams. The newly released MCP 2026-07-28 specification transitions MCP to a fully stateless protocol alongside updated SDKs across TypeScript, Python, Go, and C#. The protocol eliminates mandatory handshakes and session identifiers from core request paths while enabling servers to run on request-scoped infrastructure like Cloudflare Workers. Interactive elicitation is restructured through Multi Round-Trip Requests, which allow servers to return input requirements without sustaining persistent connections. Standardized headers for method names and improved OAuth authorization mechanisms further enable standard web infrastructure to inspect and manage MCP traffic efficiently.
Context
Early MCP implementations relied on stateful connections derived from local STDIO transports, requiring remote deployments to manage sticky sessions, open streams, and complex request routing on web infrastructure.
Approach / What changed
The MCP 2026-07-28 specification made the protocol stateless by removing required handshakes and session headers, adding Multi Round-Trip Requests for elicitations, exposing JSON-RPC metadata via HTTP headers, and updating authorization patterns.
Takeaways
- The MCP 2026-07-28 specification removes required handshake sessions and Mcp-Session-Id headers, allowing MCP servers to run statelessly on standard serverless infrastructure.
- Multi Round-Trip Requests replace open streams for elicitations by allowing servers to return an input_required status that clients answer in a follow-up retry.
- Streamable HTTP requests now include Mcp-Method and Mcp-Name headers, enabling gateways, rate limiters, and firewalls to route and inspect traffic without parsing JSON bodies.
Related reading
Secure all your internal vibe-coded applications — in one click
AI-assisted development enables employees to build applications rapidly, but unmanaged deployments can accidentally expose internal company data to the public Internet. Cloudflare introduced direct Cloudflare Access integration for Cloudflare Workers, allowing organizations to enforce authentication policies directly at the account or individual Worker level rather than per hostname. When enabled, incoming requests are authenticated before reaching application code across custom domains, routes, workers.dev subdomains, and preview environments. Developers can access authenticated identity data such as emails and groups directly via the Worker context object without manually validating JSON Web Tokens. This capability was implemented on Cloudflare's Rust-based FL2 proxy, which separated Worker routing from Worker execution so routing runs safely prior to Access policy enforcement.
Chythra MalapatiUnveiling good and bad behaviors on the Agentic Internet
Web traffic increasingly shifts between human interaction and automated agents, rendering point-in-time security checks ineffective against sophisticated automation. Cloudflare addresses this shift through continuous behavioral analysis designed to evaluate trust across entire user sessions rather than relying on static point-in-time hurdles. Its CDN-injected client-side system, Precursor, tracks continuous behavioral indicators like cursor movements to catch mid-session automated takeovers. To counter rapid bot adaptation, Cloudflare is introducing self-adjusting machine learning models under Adaptive Intelligence and non-deterministic defenses such as the AI Labyrinth. These tools allow site owners to poison scraper training data, trap crawlers in synthetic webs, or queue legitimate agents while dynamically penalizing abusive traffic.
Jin-Hee LeeFrom all-or-nothing to task-based OAuth consent
Cloudflare introduced OAuth scope customization to replace all-or-nothing consent flows for third-party applications. Previously, users had to approve every requested scope or deny access completely, compelling developers of tools like MCP servers to construct custom pre-consent selection interfaces. With the new feature, developers can designate specific permissions as optional alongside required scopes when configuring an OAuth client. During authorization, users can deselect these optional scopes, while unrequested client scopes remain hidden to maintain task-specific consent screens. Because resulting access tokens contain only the granted subset of permissions, client applications must inspect returned scopes after code exchange rather than assuming full authorization.
Miller VargasThe Cloudflare Blog – Brought to you by EmDash
Cloudflare migrated its official blog to EmDash, an Astro-focused content management system, acting as internal Customer Zero to evaluate scaling and usability. To prepare for traffic spikes reaching thousands of requests per second, engineers used k6 to execute ramp, breakpoint, and burst load tests. The resulting production architecture runs EmDash on Cloudflare Workers, pairing it with Workers Cache, an object cache backed by Workers KV, and Hyperdrive connected to PlanetScale. This layered caching design serves 99.5 percent of static files and 70 percent of requests from cache, flattening p95 latency under real-world loads reaching 850 requests per second. The migration also introduced a frontend redesign using the Kumo design system and added Model Context Protocol servers for automated agent workflows.
Kody Jackson