Model Context Protocol (MCP)

Evidence (17)

• 3Hacker News·10/1/2026product_launch
  Figma limits MCP access to approved clients, leaving Pi out

  Figma says its remote MCP server only accepts clients on its supported list, and Pi is not currently included. The company points users to its MCP catalog and a submission form for future consideration of additional clients.

• 5The New Stack·10/1/2026product_launch
  Unified API layers aim to unblock AI-assisted development

  AI agents are helping teams build applications and internal tools faster, but access to live operational data remains a major bottleneck. The article argues that a unified API layer with least-privilege, role-based access can centralize permissions, give agents their own identities, and provide one typed interface across databases and APIs via MCP and SDKs.

• 4The New Stack·9/30/2026framework_update
  Pi adds MCP support in its core

  Earendil Engineering says Pi now supports Model Context Protocol (MCP) in its core after previously rejecting it. The company says MCP has evolved and that the implementation also improves related capabilities such as Codemode and deferred tool loading. It argues the new approach makes MCP work better with structured tools and a JavaScript sandbox.

• 6InfoQ·9/25/2026framework_update
  Stateless MCP Removes Sticky Session Needs for AWS Deployments

  AWS says the latest Model Context Protocol specification removes protocol-level sessions, sticky-session requirements, and session storage for remote MCP servers. This allows requests to be routed independently and makes horizontal scaling simpler, while pushing state, retries, observability, and idempotency concerns to other layers.

• 5InfoQ·9/23/2026framework_update
  Morgan Stanley modernizes APIs for AI agents with MCP and A2A

  Jim Gough and Andreea Niculcea describe how Morgan Stanley is using Architecture as Code with CALM to update its API program for agentic systems. The approach adds Model Context Protocol (MCP), Agent-to-Agent (A2A) communication, automated governance gates, and zero-downtime upgrades to support enterprise AI at scale.

• 7Hacker News·8/22/2026framework_update
  MCP roadmap expands protocol work on messaging, transport, and identity

  The Model Context Protocol (MCP) team published an updated roadmap for the next specification release and beyond. It prioritizes agentic messaging primitives, HTTP-native transport unification, enterprise-ready agent identity and security, improved primitives, and better SDK developer experience. The roadmap also highlights ongoing work on server-initiated events, Tasks, subscriptions, progress notifications, DPoP, and related OAuth standards.

• 8The New Stack·8/8/2026framework_update
  OpenAI, AWS, GitHub and others back Agent Plugins standard

  OpenAI, AWS, Cursor, GitHub and Microsoft have backed Agent Plugins 1.0.0, an open, vendor-neutral packaging standard from Vercel for reusable AI agent extensions. The specification aims to make agent skills and MCP servers easier to distribute and load across different clients and runtimes, though some developers warn it does not solve the harder governance and permission issues.

• 7The New Stack·8/7/2026framework_update
  Coinbase, Shopify and Ramp build in-house coding agents while still using Anthropic

  Coinbase, Shopify, and Ramp have each developed internal coding agents for software engineering workflows, but they continue to rely on commercial tools and frontier models from Anthropic, OpenAI, and Google. The article argues that enterprises are increasingly owning the agent harness — the execution layer for context, permissions, orchestration, tool use, and verification — rather than the model itself.

• 5InfoQ·8/7/2026product_launch
  Instacart Launches Blueberry for AI-Assisted Incident Response

  Instacart introduced Blueberry, an AI-assisted incident response system for helping on-call engineers investigate production issues more quickly. The system combines AI agents, operational data, and past incident knowledge to produce grounded root-cause hypotheses in Slack while keeping engineers in control.

• 7InfoQ·8/7/2026framework_update
  Uno Platform 6.6 Adds Native AOT, Vulkan, and Accessibility Improvements

  Uno Platform 6.6 adds Native AOT publishing support across five target platforms, an optional Vulkan rendering backend, and automatic registration for its Model Context Protocol servers. The release also reduces XAML boilerplate, broadens WinUI API coverage, and improves accessibility and multilingual text handling.

• 7InfoQ·8/7/2026product_launch
  Azure API Management Adds Dedicated AI Gateway Tier

  Microsoft has introduced a dedicated AI Gateway tier in Azure API Management as a public preview. The new control plane is designed around models, MCP servers, and tools, and can front services such as Foundry, Bedrock, Vertex AI, and OpenAI behind a single endpoint using policy cards instead of XML.

• 7Cloudflare Blog·8/6/2026framework_update
  Cloudflare pushes an open Agentic Internet

  Cloudflare outlines a vision for an “Agentic Internet” where bots and agents can read, discover, call, and pay using open standards. The company highlights protocols such as Web Bot Auth, PACT, x402, and MCP as part of an interoperable stack for identity, access, and payments.

• 7Cloudflare Blog·8/6/2026product_launch
  Cloudflare adds agent readiness and AEO tools for AI assistants

  Cloudflare introduced Agent Readiness diagnostics and an Answer Engine Optimization (AEO) tool in its dashboard to help site owners understand how AI agents access, read, and recommend their sites. The tools evaluate signals such as robots.txt, sitemaps, Markdown output, metadata, and emerging agent standards, then provide recommendations and setup links to improve agent compatibility.

• 7InfoQ·8/6/2026product_launch
  Cloudflare previews WebMCP support for websites

  Cloudflare has launched a developer preview of WebMCP that lets site owners expose browser-accessible tools to AI agents with a simple toggle, without changing origin code. The implementation adds an edge-side bridge and supports tool packs such as Content Credentials and a Site MCP Server pack that can work entirely in the visitor’s browser.

• 7Cloudflare Blog·8/6/2026product_launch
  Cloudflare adds managed AI Search for agents

  Cloudflare announced developer experience improvements to AI Search, letting users index data, search across multiple sources, and expose public or private search endpoints with less manual setup. The update also introduces preview pricing, custom domains for endpoints, and semantic search support for sites built on EmDash.

• 6Cloudflare Blog·8/5/2026product_launch
  Cloudflare adds WriteGuard for MCP server write controls

  Cloudflare introduced WriteGuard, a shared policy, attribution, and auditing layer for its internal MCP servers. The system can allow, enrich, or block tool calls to improve control over agent write actions and make agent activity easier to trace. It is now being extended to Cloudflare MCP server portals through a private beta.

• 2Simon Willison·8/2/2026open_source
  Simon Willison publishes July 2026 newsletter preview

  Simon Willison posted an update announcing his sponsors-only monthly newsletter, with a preview of the June edition and a list of topics covered. The note highlights recent discussions of model behavior, several new model releases, and renewed interest in MCP, but it is primarily a newsletter announcement rather than a major product or research release.