Is this you? As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.
Claim your profile
Articles by Blair Hayes
Google Adds Scanning and Policy Controls for Agent Registry Skills
Google has updated the Agent Registry, a central catalog to register, discover, and govern AI agents, MCP servers, tools, endpoints, and standalone skills. According to the Nov. 6 release notes, standalone skills have reached general availability. It is critical to distinguish that while these skills are GA, the newly introduced security scanning and governance capabilities are available for use but are not separately labeled as GA.
NVIDIA Leadership Transition and Regulatory Inquiries: A Status Report Original
NVIDIA has entered a period of significant leadership transition following the immediate resignation of Jensen Huang from his roles as president, CEO, and board member. The company has appointed CFO Colette Kress as interim CEO, while Harvey Jones has been designated as the first board chair. The official company announcement provided no specific reason for Huang’s departure.
Goose Implements Deferred MCP Tool Schema Loading
Goose has integrated a new mechanism for handling Model Context Protocol (MCP) tools, targeting how their schemas are presented to the model. The implementation, detailed in Goose PR #13378, adds a deferred loading strategy. This story concerns the merged Goose code change; it does not establish a separate tagged product release or a change to the MCP protocol itself. By default, Goose keeps its existing eager behavior, where tool schemas are provided to the model. Deferral is opt-in.
OpenAI Moves Research Stop Enforcement Into the Inference Path
OpenAI’s Control Room design places stop enforcement for research workloads in the inference path. The system tracks model workloads across its research infrastructure, including agentic sampling for model training and evaluation, data processing and more. It also works with misalignment monitoring. Calling it a control plane is descriptive: the term captures its authorization role, not OpenAI’s official name.
OpenAI Agents SDK alpha changes three configuration surfaces
OpenAI Agents SDK v0.24.0-alpha.7 is labeled a pre-release in the official release notes. The GitHub release API marks it as a prerelease and records publication at 2026-11-08T07:05:11Z in the release API record. The distinction matters: this is an account of what the alpha specifies, not a general-availability launch or evidence of adoption. Its practical significance for builders is narrower and more useful: one version changes assumptions across three separate configuration surfaces.
Google ADK v2.14.0 replaces shell-command input for skill scripts
Google ADK v2.14.0 changes how RunSkillScriptTool is called when it runs in environment mode with a SkillToolset. The tool no longer accepts a free-form shell command string; callers instead provide args, short_options and positional_args. The scope is specific: the release notes describe this contract for that environment-backed configuration, not a change to every ADK tool or every way a script might be run. The interface change is one part of the update.
Cloudflare frames agent identity as a site-governance problem, not just a bot-control setting
When a remotely hosted agent acts as an intermediary between a user and a website, the site often interacts with the agent rather than the human user. This shift creates a fundamental governance challenge: how to recognize, serve, and hold operators accountable while maintaining privacy. To navigate this, operators must distinguish between three distinct layers of agent identity: access permission, agent or operator verification, and user or account attribution.
WebMCP’s Cross-Origin Tools Need Three Opt-Ins. Cloudflare’s Backends Differ
A draft can define a careful cross-origin gate without making every browser use that same gate. The WebMCP Community Group draft, dated November 2, 2026, proposes a way for web pages to expose structured tools. It is not a W3C Standard and is not on the W3C Standards Track. For builders, the key distinction is between the draft’s in-page JavaScript access model and the separate way a browser agent may inspect tools.
Understanding the v2.3.0 MCP Server Release and Its Granular Controls
A Modular Approach to Server Configuration The official GitHub release of the Model Context Protocol (MCP) servers, versioned as v2.3.0 and published on November 4, 2026, introduces targeted operational controls and Git tools. Rather than a blanket lockdown, it adds opt-in filesystem and fetch settings alongside separate Git operations. Operators should be aware of the package versioning landscape in this release.
MCP Inspector 2.14.0 fixes three privately reported vulnerabilities and changes developer-facing behavior
MCP Inspector 2.14.0 was timestamped at 23:55:33 UTC on November 4 in the project’s release feed. The 2.14.0 release notes say the update fixes three vulnerabilities reported privately against v2. They identify one of the three, GHSA-2fw8-7jmw-r937, as also affecting the deprecated v1 line; the release notes point to v1.0.4 as the fix there. The page does not assign severity or give a broader affected-version range.
OpenAI Agent Builder’s Nov. 30 shutdown puts migration on a code-and-validation path
On June 3, 2026, OpenAI announced that Agent Builder is scheduled to shut down on November 30, 2026. For teams that choose to carry workflows forward, the practical question is which documented route fits. The official deprecations page says the product is being retired and that ChatKit remains available.
Google Bundles Cloud Docs and Guardrails Into an Antigravity Plugin
On November 2, 2026, Google released the google-cloud-developer plugin, a new tool designed to integrate cloud-management capabilities directly into agentic workflows. Available through the Build with Google section of the Antigravity Marketplace, the plugin can be installed by running agy plugin install google-cloud-developer.
What Apify’s TOON Trial Can (and Can’t) Tell MCP Builders
Apify’s 30-day account of TOON in its Model Context Protocol (MCP) server makes a seemingly narrow change harder to judge: the proposal, release and follow-up tests did not examine the same intervention. The result is a case study in how to test an encoding choice, not a verdict on TOON across MCP. Apify says it ran TOON in production on five tools, then removed it after two live A/B rounds showed no consistent benefit. The first comparison was a proposal, not a live model evaluation.
Android AppFunctions Beta Does Not Mean Open Agent Access
AppFunctions beta and production publishing The October 28, 2026 Android Developers Blog announcement says AppFunctions is now in beta. It describes AppFunctions as Android’s unified tool-calling layer and says app capabilities can be shared with agents through the Model Context Protocol (MCP). The same announcement says developers must sign up for the Connected Apps Platform preview to publish to production across surfaces. Beta API status and production publishing are distinct.
Pydantic AI v2.64.0 Adds Sticky Model Fallback for Conversation Continuity
The release of Pydantic AI v2.64.0 on October 30, 2026, introduces an opt-in sticky setting for ModelFallback, as detailed in the official release notes and the associated pull request #11330. This feature allows a conversation to remain on a fallback model that has successfully answered a request, rather than defaulting back to the primary model at the start of every subsequent request step.
GitLab Orbit Puts a Shared Lifecycle Graph Behind Multiple Agents
GitLab Orbit provides a live, queryable graph that encompasses the software lifecycle, including code, merge requests, work items, pipelines, vulnerabilities, and ownership. As detailed in the official announcement, GitLab describes the platform as a unified data layer designed to standardize how various agents interact with organizational context. Architecturally, GitLab characterizes Orbit as a shared layer accessible by multiple clients.
Agents Choose When; Step Functions Keeps the Workflow’s Steps
AWS has made existing Standard and Express Step Functions workflows callable as tools through Amazon Bedrock AgentCore Gateway, an agent gateway. The key design boundary is simple but consequential: an agent chooses whether to invoke a workflow and what input to pass; the configured state machine determines which steps run and in what order. The connector’s two jobs In its October 29 announcement, AWS describes using existing Standard and Express workflows as agent tools.
Google Cloud Puts Agent and MCP-Server Traffic on a Static Egress Path: VPC Connectivity Goes GA
Google Cloud’s release notes carry an entry dated October 27, 2026: VPC connectivity and static IP egress for custom MCP servers is generally available as a Gemini Enterprise feature. The GA covers a specific set of networking capabilities, currently supported in the us-central1 and europe-west1 regions. It is not a blanket announcement about the status of the broader Agent Gateway service. What it changes is where enterprise control over agent behavior can be exercised.
Claude Ships MCP Tasks and Elicitations in Claude Code — Durable Agent Workflows, Narrow Launch
Anthropic shipped support for MCP Tasks and Elicitations in the Claude Code CLI on October 23, 2026 — the client-side move that turns Model Context Protocol (MCP) tools from single request/response calls into durable, interruptible work. David Soria Parra, the MCP co-creator at Anthropic, previewed the feature in his keynote at AGNTCon & MCPCon North America on October 22; the shipping record is the official developer thread the following day.
Grok Bot Will Route Your Tasks to Claude. The Moat Just Moved to the Switchboard.
The Interface-Layer Moat On October 7 at 06:46 UTC, Elon Musk posted: “Going forward, SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs.” He added: “Whatever is most likely to give you the best outcome.” Competitor CellCog, which builds AI employees in the same category and discloses the conflict, reported the post passed 30 million views in seven hours.
AAIF Portfolio Doubles as Four Labs Donate Core Agent Rails
The Agentic AI Foundation (AAIF) has reached a significant milestone in its first year, expanding its portfolio to 10 hosted projects. This growth, up from the three projects present at the foundation’s December 2025 launch, signals a shift toward centralized governance for agentic infrastructure. The expansion is driven by a coordinated donation act from four major labs: Anthropic, OpenAI, Hugging Face, and Alibaba. The four new additions cover critical layers of the agent stack.
MacBook Pro with M5 Pro Trails NVIDIA RTX Spark N1X in Local Agent Benchmark.
NVIDIA’s new RTX Spark N1X platform has demonstrated a significant performance lead over Apple’s M5 Pro silicon in independent local AI agent testing, marking a shift in the competitive landscape for on-device inference hardware. The RTX Spark N1X laptop, which saw a worldwide release on October 16, 2026, was evaluated against the 14-inch MacBook Pro (M5 Pro, 64 GB) in the AA-AgentPerf-Local leaderboard .
Git Breaks at Agent Scale. Anthropic’s Replacement Took Five Months and Ships to No One.
In May 2026, Anthropic averaged roughly 2,500 code changes merged per day. By September, that figure had climbed to over 30,000 per day, a 12-fold increase in four months, according to company-reported data. That surge is a structural breaking point, and the company names what broke in its redesigning source control post: coding is no longer the bottleneck, and the constraint is the developer infrastructure built for human scale, namely source control, builds, and CI.
NVIDIA Kumo: Open-Weight Models for Reasoning Over Structured Enterprise Data
On October 21, 2026, at GTC Berlin, NVIDIA announced the Kumo family of foundation models, designed specifically to connect AI agents to structured enterprise data. Part of the broader Agent Knowledge Foundation, Kumo addresses the ‘Data Plumbing’ challenge identified in the TGVP infrastructure report, where the inability to reason over databases remains a primary barrier to autonomous agent deployment.
The AGNTCon Ledger: What Shipped This Week for Agents and What Is Still a Slide
The loudest story in agent infrastructure as AGNTCon opens this week is not what arrived, but what remains missing from the ledger. On October 22, 2026, as AGNTCon and MCPCon open in San Jose, the industry faces a fundamental gap: while individual governance stacks are shipping, the connective tissue that would make them interoperable remains a series of drafts and blueprints.
NVIDIA Pushes Agent Enforcement Into the Silicon, and Publishes the Blueprint
NVIDIA announced the OpenShell Reference Architecture at GTC Berlin today, the third act in a fast sequence: the OpenShell runtime shipped September 16, the Open Agent Safety Platform followed September 28, and today’s release turns that platform into a published, full-stack blueprint for industrial-scale agent operations. The stated target is the primary bottleneck to enterprise agent adoption: governance, pushed down from the application layer into the hardware. The architecture has four legs.
Proof and Circle Demonstrate End-to-End Agentic Commerce: The x401/x402 Handshake Original
While traditional financial institutions remain locked in a cycle of blocking agentic traffic, the open-protocol ecosystem is demonstrating a functional alternative. On October 19, 2026, at Money20/20 USA, Proof and Circle jointly unveiled an end-to-end demo of an AI agent completing an identity-gated merchant purchase. This ‘Protocol Handshake’ bypasses the bank blockouts currently stalling the agent economy by replacing institutional gatekeeping with cryptographic verification.
AWS and Google Both Registered Their Chips Through One PyTorch Interface in a Day. The Lock-In Moves Up the Stack.
At the PyTorch Conference North America in San Jose, Amazon and Google both registered their proprietary silicon portfolios through the same PyTorch extension point within roughly one day. Interface portability is now table stakes for hardware vendors, even as the underlying competition for performance and supply remains as bilateral as ever. Both backends are released under the Apache 2.0 license and integrate with the TorchTitan and Hugging Face libraries.
Rhapsody Guardian: Healthcare Builds Its Own Agent Gatekeeper, Ahead of Any Shared Spec.
On October 20, 2026, at the Rhapsody Connect Executive Track in Boston, Rhapsody announced the general availability of Guardian, a healthcare-native gatekeeper for AI agent access to clinical data and services. It evolves the company’s earlier API Guardian, introduced in October 2025 – a distinct, earlier product – from static API management into runtime agent governance. The release’s shortest useful line: Guardian is available now. This is a shipped control plane, not a roadmap.
Rhapsody Ships Guardian: Healthcare Builds Its Own Agent Gatekeeper Ahead of Any Shared Spec
On October 20, 2026, at the Rhapsody Connect Executive Track in Boston, the company announced the general availability of Guardian. This release marks a significant evolution from the company’s earlier API Guardian, introduced in October 2025, shifting the focus from static API management to dynamic, runtime agent governance. Healthcare organizations face a persistent mismatch between the open-ended nature of AI models and the strict requirements of clinical environments.
Google and Meta Open-Source TorchTPU: Decoupling Model Code from Silicon
Google and Meta officially open-sourced TorchTPU at the PyTorch Conference North America in San Jose on October 20, 2026. While this release changes how developers interact with specialized hardware, the underlying engineering was not developed for this announcement. Google published its deep dive, TorchTPU deep dive, on April 7, 2026. The code arrived today, but the technical foundation has been maturing for months.
CAEVES Ships Context LAKE: Enterprise Context as a Managed MCP Service
The current state of enterprise AI is defined by a persistent gap: agents can reach the world, but they cannot reliably access the specific, siloed enterprise data that would make them useful. While models have become increasingly capable, the infrastructure required to bridge them to internal knowledge remains fragmented. On October 7, 2026, CAEVES Technology, Inc. announced a PR Newswire release for Context LAKE, a platform designed to serve as an enterprise context layer for AI assistants.
Vercel Built a Durable Agent State Stack. The Reference App Is Open-Source and Running on AI Gateway.
Vercel’s internal deployment data reveals a sharp shift in how developers are utilizing their platform: coding agents have surged from under 3% to more than 50% of total deployments in just six months. This transition is underscored by a 28x increase in agent-triggered deployments, signaling that autonomous workflows are moving from experimental scripts to core production components. However, this rapid adoption has exposed significant fragility in existing systems.
Okta Advances Agent Identity Framework as Market Builds Around Pre-GA Standards
The enterprise transition to AI agents is hitting a fundamental infrastructure wall: the reliance on static, long-lived API keys. These credentials, often stored in plain text or configuration files, represent a significant security liability. A new approach is emerging to replace these standing permissions with identity-governed, short-lived tokens issued at runtime. This shift represents a move toward treating agents as first-class identities rather than generic machine principals.
Google Handed Its Agent Framework to an Agent. The 2-of-6 Review Ratio Is the Real Story.
When Gemini 4 Argon spent 11 hours overnight optimizing Google’s Agent Development Kit, it didn’t just produce a performance boost; it provided a masterclass in the current limits of autonomous engineering. The primary hurdle for enterprise agent adoption is not model intelligence, but the maturity of the tooling layer – one of four bottleneck layers identified by TGVP’s infrastructure report.
Accenture Federal Services and Google Public Sector Deploy Gemini Launchpads for Federal AI Agents
Federal agencies are currently struggling to reconcile the rapid evolution of autonomous AI with the rigid, non-negotiable compliance requirements of the federal technology stack. While frontier models have generated significant interest, the transition from experimental pilots to mission-critical deployment is stalled by infrastructure constraints.
Bifrost and Cloudflare Both Replaced Tool Catalogs with Code Generation. The Agent Tooling Layer Is Standardizing on a Pattern Nobody Named.
Two independent infrastructure vendors arrived at the same architectural conclusion within months of each other, and the agent tooling layer is quietly standardizing on a pattern that nobody has formally named. Bifrost, the open-source AI gateway from Maxim AI, and Cloudflare, through its Agents SDK, both replaced large MCP tool catalogs with direct code generation. The convergence is not a coincidence.
Solo.io Agentgateway Joins AAIF — The Agent Gateway Layer Gets a Governance Home
The infrastructure supporting agentic AI is undergoing a structural shift. On June 4, 2026, Solo.io’s open-source agent gateway became the fourth initiative hosted by the Linux Foundation’s Agentic AI Foundation (AAIF).
TGVP: Infrastructure, Not Model Quality, Is the Primary Bottleneck to Enterprise AI Agent Adoption
Enterprise AI adoption is currently characterized by a significant gap between pilot programs and production-grade deployment. According to data cited by the TGVP report, published April 29, 2026, by Vignesh Radhakrishnan, 79% of companies are actively adopting AI agents, yet only 2% have managed to deploy them at scale, as reported by Capgemini. This chasm between intent and execution is the defining tension of the current market.
China Proposes Export Controls on 23 AI Technologies Including AI Agents
China’s Ministry of Commerce and Ministry of Science and Technology issued Announcement No. 48 of 2026 on October 16, signaling a fundamental shift in the nation’s approach to AI governance. For the first time, Beijing is proposing outbound export controls on AI technologies. While previous regulatory efforts focused on restricting inbound access to foreign models or managing domestic agent behavior, this proposal targets the export of Chinese AI intellectual property and infrastructure.
China Proposes Export Controls on 23 AI Technologies Including AI Agents
China Proposes Export Controls on 23 AI Technologies Including AI Agents Sun, October 18, 2026 at 10:07 AM EDT China's Ministry of Commerce and Ministry of Science and Technology issued Announcement No. 48 of 2026 on October 16, signaling a fundamental shift in the nation's approach to AI governance. For the first time, Beijing is proposing outbound export controls on AI technologies.
The Industry’s Primary Agent Identity Standard Expires in 36 Days
The clock is ticking on the foundational protocol for Agent SSO. As of October 17, 2026, exactly 36 days remain before the current revision of the ID-JAG draft expires on November 22, 2026. While draft expiration is a routine administrative milestone in the IETF lifecycle, this one highlights the widening chasm between commercial agent infrastructure that is shipping now and the institutional process that is supposed to anchor it.
Six Vendors Are Shipping Agent Identity Products. The IETF Has Not Finished the Standard.
Six vendors are now shipping agent identity products. The IETF has not finished writing the standard any of them would need. Our coverage of the IETF standardization gap mapped the structural problem: ID-JAG remains an Internet-Draft (-04, expires November 22, 2026), the agentproto Working Group has a contested charter, and the WIMSE and DAWN efforts are either scoped away from agents or still early. What that piece did not map is what deployers actually have to choose from right now.
The IETF Is Formalizing the Identity Layer Vendors Are Already Shipping Original
The rapid proliferation of autonomous agents has created a significant governance vacuum. As organizations prepare to manage an estimated 150,000 agents by 2028, the technical infrastructure required to secure these entities is being built in real-time. While the Internet Engineering Task Force (IETF) works to formalize the necessary identity layers, the market is not waiting for consensus.
Commoditized: The Agent Sandbox Layer Has Become a Utility — and the Moat Is Moving
Autonomous agents are moving from experimental sandboxes to production-grade infrastructure, forcing a structural shift in how we isolate execution. The commoditization of per-agent isolation has moved from niche experimentation to a baseline requirement — and a multi-vendor convergence is now visible across the stack. This is not merely about the proliferation of sandboxes. It is about the fundamental decoupling of execution from the control plane.
Perplexity Ships Its Search Agent’s Sandbox as Open Source — and the Sandbox Layer Just Got Another Entrant
On October 15, 2026, Perplexity released its execution sandbox to the public. The SPACE repository — Sandboxed Platform for Agentic Code Execution, written in Rust — carries an Apache 2.0 license, 820 stars as of October 16 (770 in the first 24 hours per the announcement), and the production record of powering every Perplexity Computer agent session since July 2026.
Identity as the Moat: Descope Embeds MCP-Native Auth into the Gateway Layer
The infrastructure supporting autonomous agents is undergoing a fundamental shift as identity verification moves from a peripheral security check to a core gateway component. With the release of Agentic Identity Hub 3.0 on October 15, Descope has become the first Non-Human Identity (NHI) vendor to embed Model Context Protocol (MCP) native authorization directly into the gateway layer.
Fifteen Days Out: AWS Retires the Namespace Its Agent Registry Was Born In
On October 30, 2026, AWS discontinues support for the namespace the AWS Agent Registry was born in. Teams have fifteen days from today, October 15, to complete their migration to the new agent-registry namespace. On the public record this desk tracks, no major cloud has retired an agent registry namespace and forced a customer data migration before; a namespace retirement is not a deprecation warning, it is a structural change to where agents are named and found.
Eight Days Out: AGNTCon Keynote Assignments Map the Agent Control-Plane Playbook
The AGNTCon+MCPCon North America 2026 keynote schedule and the program’s Featured Keynote Speakers page function as a strategic map of the emerging agent infrastructure landscape. By analyzing the speaker assignments, we can identify how major vendors are positioning their platforms and where they are concentrating their influence. The stage in San Jose this October reveals a clear divide between those prioritizing broad platform governance and those focusing on specialized technical foundations.
The Biggest API Platform Just Bet Its Future on Protocol-Agnostic Agent Governance
On September 29, 2026, Postman shipped Fabric Gateway. The largest API platform in the world just bet its future on protocol-agnostic agent governance as a control plane. Fabric Gateway is a hybrid deployment: Postman manages the control plane; the customer runs the data plane inside their own VPC. It sits between agents and everything they touch—HTTP endpoints, gRPC services, MCP servers, and A2A agents—and enforces policy uniformly across all of them.
Show More
loading
Actions
Is this you?
As a journalist, you can create a free Muck Rack account to customize your profile, list your contact preferences, and upload a portfolio of your best work.Get in touch with Blair
Contact Blair, search articles and posts on X, monitor coverage, and track replies from one place.
Learn more about Muck Rack