---

OpenAI and authentication specialist Yubico have teamed up to offer custom hardware security keys that will allow users to lock down access to their ChatGPT accounts. The method uses hardware-backed passkeys that store login credentials on a physical device rather than relying on passwords that can be stolen through phishing attacks or reused across services. The new offering includes a two-pack of custom YubiKeys tied to OpenAI’s Advanced Account Security program, giving users physical authentication devices instead of relying only on software-based verification. Yubico hardware Each set includes a YubiKey C NFC, which supports tap-to-authenticate on phones, and a YubiKey… [Continue Reading]

Disclaimer: This article was created using AI-based writing and communication companions. With its help, the core topics of this rich and nuanced livestream were distilled into a compact blog post format.

In our recent JetBrains livestream, Vitaly Bragilevsky was joined by Rain, the creator of cargo-nextest, for a conversation about Rust testing, developer tools, open source maintenance, and the everyday developer experience around large Rust projects.

cargo-nextest is widely used across the Rust ecosystem as a next-generation test runner. It is built to make Rust test execution faster, more observable, and more reliable, especially for larger codebases, CI pipelines, and projects with complex integration tests.

The conversation also came at a good time for RustRover users. RustRover 2026.1 introduced native cargo-nextest support, so developers can now run and monitor nextest sessions directly from the IDE, with progress reporting and structured results in the Test tool window.

If you missed the livestream, you can watch the full recording on JetBrains TV. Below, you’ll find a structured recap of the key questions and insights from the session.

Q1. Who is Rain, and how did they get into Rust?

Rain is a software engineer with more than a decade of industry experience. Their Rust journey started professionally in 2017 while working on source control infrastructure at Meta. Rain joined a project to build a Mercurial server in Rust. The team already had someone with Rust expertise, while Rain brought deep knowledge of Mercurial. That collaboration became their entry point into Rust.

They have worked at Mozilla, Meta, and now Oxide Computer Company, where Rust is used throughout the stack, from embedded firmware to higher-level control plane software.

Q2. What is cargo-nextest?

At a high level, it is designed to run Rust tests faster and give developers better insight into what happened during a test run. In benchmarks,cargo-nextest can be up to three times faster than cargo test, depending on the project and workload.

cargo-nextest also includes features that become important as projects grow. That combination is what makes cargo-nextest useful across both open source and large industry codebases.

Q3. What problem does cargo-nextest solve?

Rain was clear that cargo test is still a good tool, especially for testing core algorithms, data structures, and smaller projects. The limitations become more visible when a Rust project grows into a large service, large CLI application, or codebase with many integration and end-to-end tests.

In those cases, the main problems are not only “how fast can I run tests?” but also:

• Which tests are slow?
• Which tests are flaky?
• What happened in CI?

cargo-nextest is built for that kind of environment.

Q4. When should someone switch from cargo test to cargo-nextest?

If you are happy with cargo test, you do not have to switch. But if you are unhappy with some part of the testing experience, cargo-nextest is worth trying.

They pointed to three common signals.

1. Flaky tests. cargo-nextest can retry tests, which helps distinguish between a consistent failure and a flaky one. 
2. Test isolation. cargo-nextest runs every test in its own process. This matters for tests that rely on global state, external APIs, graphics contexts, or other resources that may not behave well when reused across tests. 
3. Speed. For large services and bigger test suites, cargo-nextest is often several times faster than cargo test. That can save both local developer time and CI compute.

Q5. What is the coolest cargo-nextest feature?

With run recording, cargo-nextest can capture what happened during a test run, including in CI. Developers can then fetch that run locally, inspect what each test did, and better understand failures that happened outside their machine.

Second one is Perfetto trace output. cargo-nextest can generate trace data that can be opened in Perfetto, giving developers a graphical view of test execution. 

Q6. What lesser-known feature should more people know about?

When developers debug tests with cargo test, they often end up running the test binary directly. The problem is that Cargo normally sets up environment variables and configuration around test execution.

cargo-nextest’s debugger support helps solve that by configuring the environment so the test runs in the debugger in a way that is equivalent to normal execution.

For developers who need to step through a single failing test, that can make the difference between chasing a misleading local reproduction and actually debugging the real failure.

Q7. How does cargo-nextest help with stuck or long-running tests?

cargo-nextest lets you dump information about currently running tests. On any operating system, you can press T; on macOS, you can also use Control-T via SIGINFO. This gives you a live view of how long tests have been running, along with stdout and stderr output.

That is especially helpful for debugging complex failures that only appear in the context of a larger run.

Q8. What was the hardest part of building cargo-nextest?

A test runner has to do much more than start a process and wait for a result. It has to observe what happens, handle success, failure, abnormal termination, timeouts, retries, output capture, scheduling, and more. As new features are added, that state machine becomes more complex.

Interestingly, cargo-nextest is not a classic “10 million requests per second” async use case. It usually runs only as many tests as there are CPU cores or threads. But the state machine itself is complex, and async Rust gives a structured way to manage that complexity.

Q9. What would Rain change about Rust if they could go back to 2017?

Async Rust is powerful, but it also introduces footguns that are not present in the same way in synchronous Rust. Rust’s safety model is one of the reasons they were attracted to the language, especially around thread safety, aliasing, and mutability. Async Rust keeps many of those strengths, but cancellation and cleanup can be difficult to reason about.

Q10. Is cargo-nextest extensible?

Yes, and this is one of the reasons it has been able to integrate with other tools.

cargo-nextest provides machine-readable output formats and extension points that other tools can build on. It also supports features like setup scripts, which can prepare a database, seed test data, or configure an environment before tests run.

Q11. Can cargo-nextest be useful for embedded Rust?

Cargo-nextest itself probably will not run on embedded hardware because it is fairly complex. But it can still help with embedded testing workflows where tests are dispatched to real hardware.

One relevant feature is wrapper scripts. Instead of executing a test directly, cargo-nextest can run a script around it. That script can set up the environment, send commands to the hardware, or coordinate with a target runner.

Q12. What makes a great developer tool?

For Rain, a great developer tool has to be correct and reliable, especially when things go wrong.

For a test runner, that means handling more than the happy path. Tests can fail, time out, crash, segfault, produce output, or behave differently depending on the environment. A good tool needs to represent all of that clearly.

That is a useful principle far beyond test runners. A great developer tool does not try to remove the developer from the work. It helps them do the work with more confidence.

Q13. Which Rust tools does Rain recommend?

Rain mentioned three:

• cargo-hack: Useful for crates with many feature combinations. It can run tests across feature sets and optimize combinations based on dependencies between features.
• cargo-expand: Useful when developing macros or procedural macros because it shows the expanded output.
• cargo-semver-checks: Useful for crate maintainers because it detects semver compatibility issues, including simple API changes and more subtle changes such as accidentally removing Send or Sync bounds.

Good Rust tooling helps developers understand what their code is doing, what changed, and what might break.

cargo-nextest in RustRover

RustRover 2026.1 adds native support for cargo-nextest directly in the IDE. For Rust developers, this means you can run and monitor nextest sessions without leaving your normal development workflow.

Instead of switching to the terminal and reading raw output, you get structured test results and progress reporting in the Test tool window.

The goal is not to replace the terminal for developers who prefer it. The goal is to reduce friction for teams and developers who already rely on cargo-nextest and want the same workflow integrated into the IDE.

Closing thoughts

As Rust projects grow, testing becomes part of how teams understand correctness, performance, and confidence.

cargo-nextest helps make test runs faster, more isolated, more observable, and easier to debug.

If you’re interested:

• Explore cargo-nextest
• Try cargo-nextest support in RustRover 2026.1
• Watch the full livestream with Rain
  

If you’re working with slow test runs, flaky tests, or complex CI failures, cargo-nextest is worth exploring.

Choose an ecosystem partner for agent security and governance

AI agents aren’t coming—they’re already in your environment. They show up in places you expect (like Microsoft Copilot, Microsoft Teams, and Microsoft 365) and even more places as technology evolves (a local autonomous personal AI assistant or a new software as a service (SaaS) agent connected to your sensitive data.)

The problem isn’t that agents exist. It’s that they proliferate fast, span apps, endpoints and cloud, and often operate outside the visibility and control of the teams accountable for risk. When an agent can invoke tools, access data, and interact with other agents, any “helpful” workflow can turn into data oversharing, tool misuse, or over-privileged actions in seconds. And as agents become even easier to create and deploy, your attack surface grows with them. 

That’s why end-to-end observability matters: you can’t govern what you can’t see, and you can’t secure what you don’t understand—especially when the number of agents is a moving target. 

Microsoft Agent 365 helps you take control of agent sprawl as your control plane to observe, govern, and secure agents and their interactions—including agents built with Microsoft AI and agents from our ecosystem partners—using the admin and security workflows your teams already run. 

General availability starts today for Agent 365.

Additionally, we’re announcing the previews of new Agent 365 capabilities and integrations to help you scale agent adoption with the right controls in place. 

• Observability, governance, and security for agents operating independently—Agent 365 is expanding to cover agents that operate with their own credentials and permissions.
• Discovery of agents and shadow AI, using capabilities of Microsoft Defender and Microsoft Intune for both local and cloud agents.
• A secured, managed environment for agents to work in Windows 365 for Agents.
• Coverage for a wide ecosystem of SaaS agents, including agents innovated by software development companies (SDCs).
• Support for evaluation, adoption, and usage from Microsoft and ecosystem partners worldwide.

Manage agents with a single control plane, regardless of how or where they work

As organizations move from pilot to adoption, AI agents are being deployed across increasingly diverse use cases. Some act with delegated access, working on behalf of users; others operate with their own credentials and permissions, participating in team workflows or operating behind the scenes. 

With Agent 365, you can observe, govern, and secure AI agents whether they act on behalf of users with delegated access—for example, an agent that helps employees organize their inbox—or agents that operate with their own access and scope of work—such as an agent autonomously triaging support tickets. 

	Supported by Agent 365
Agents working on behalf of users (delegated access)  Generally available	Agents operating behind the scenes (own access)  Generally available	Agents participating in team workflows (own access)  Public Preview

Discover and manage local and cloud-hosted agents 

Users are installing agents like OpenClaw and Claude Code on their devices and adopting SaaS agents built by developers on new and emerging platforms. Many of these local and cloud-hosted agents run unmanaged and outside of traditional governance, as they autonomously execute tasks, modify code, or access confidential information, creating a new wave of shadow AI.  

To help organizations address accelerating agent sprawl and the rise of unmanaged agents, we’re introducing new capabilities as part of Agent 365, Microsoft Defender, and Intune so you can discover shadow agents, and apply appropriate controls, such as blocking unmanaged agents. 

Discover and manage local agents

With Microsoft Defender and Intune, organizations will be able to discover and manage local AI agents running on Windows devices, starting with OpenClaw agents and expanding soon to other widely used agents like GitHub Copilot CLI and Claude Code. Customers enrolled in the Frontier program can see if OpenClaw agents are being used in the organization, which devices they are running on, and use Intune policies to block common ways that OpenClaw runs on the new Shadow AI page in Agent 365 in the Microsoft 365 admin center and in the Intune admin center. Through Agent 365 registry, the inventory of local agents will be available in Defender and Intune so IT, endpoint management, and security teams can get a consistent view of discovered local agents in their environment and take appropriate action.

Starting in June 2026, Microsoft Defender will also provide asset context mapping for each agent including the devices they run on, MCP servers configured for those agents, the identities associated with them, and the cloud resources those identities can reach. This will give security teams the context needed to assess exposure and potential blast radius. They can then investigate agent activity, such as file access and network behavior, using familiar endpoint data, and use those insights to identify misconfigurations and even define custom detections.

Beyond monitoring, organizations will be able to apply policy-based controls to set guardrails for what agents are allowed to do—helping protect both agents and organizations from compromise and misuse—with initial support delivered for OpenClaw through Intune. If a managed agent exhibits malicious behavior patterns, such as attempting to access or exfiltrate sensitive data, Defender will be able to block coding agents in runtime and generate alerts with rich incident context to support investigation and response.  

Context mapping capabilities, policy-based controls, plus runtime blocking and alerts will be available in Agent 365 through Intune and Defender public preview in June 2026. 

Visibility across clouds and AI-builder platforms

As developers are rapidly building agents with Microsoft Foundry, AWS Bedrock, and Google Gemini Enterprise Agent Platform (formerly Google Vertex AI) and deploying cloud agents across multicloud and multi-platform environments, the agent sprawl challenge intensifies. To manage potential security risks or vulnerabilities before they become breaches, security and IT teams need visibility to which cloud agents are running, what models these agents are built on, and what resources they’re accessing.

Today, we are excited to announce the public preview of Agent 365 registry sync with AWS Bedrock and Google Cloud connections, using IT teams to automatically discover, inventory, and, soon, perform basic lifecycle governance—for example, start, stop, delete agents—across these platforms.

Manage a wide ecosystem of SaaS agents 

Agent 365 works with prebuilt agents in Microsoft 365 Copilot and Teams, agents built with Microsoft Copilot Studio or Microsoft Foundry for your organization, and agents built by software development companies partnered with Microsoft.

Delivering on our promise of control plane for the broad agent ecosystem, we’re excited to announce ecosystem partner agents fully configured to be managed by Agent 365, including Genspark, Zensai, Egnyte, and Zendesk, and agents built on agent factories, including Kasisto, Kore, and n8n. Organizations can observe, govern, and secure these agents in the Agent 365 control plane, with no integration work by IT or security teams.  

Agent 365 software development company launch partners

Enterprises can easily build AI agents today, but scaling them with trust and governance is where most initiatives stall. With Kore.ai deeply integrated into Microsoft Agent 365, identity, security, and governance are built in from the start—empowering enterprises to move from pilots to AI at scale with confidence.

—– Raj Koneru, Chief Executive Officer of Kore.ai

The Agent 365 developer and ecosystem partners play a critical role in extending agents into line-of-business systems, building vertical and scenario-specific integrations, modernizing legacy automation into agent workflows, extending Copilot experiences with custom agents, and helping customers operationalize agent ecosystems at scale. These Agent 365 enabled agents are then observable, governable, and securable in the Agent 365 control plane, accelerating adoption for your organization.

• Explore the Agent Showcase of ecosystem partner agents that work seamlessly with Agent 365.
• Get started with Agent 365 development with guidance from Microsoft Learn.

Secure agents as they work in Windows 365 

While Agent 365 provides the control plane to observe, govern, and secure agent activity across the enterprise, Windows 365 for Agents—now available in public preview (in the United States only)—provides a secured, managed environment where agents can carry out that work. It introduces a new class of Cloud PCs purpose-built for agentic workloads and managed in Intune, allowing agents to run in policy-controlled environments, interact with applications, and operate with the same identity, security, and management controls already used for employees.

Now, with Agent 365, you can also observe and secure agents running on Windows 365 for Agents in Microsoft 365 admin center, understanding which agents are connected to the cloud-powered compute. Together, they enable organizations to move from visibility and governance of agents to confidently running them in production environments. 

Secure agents against internet threats with network controls  

AI agents can operate much faster than human users. Without proper guardrails, they can connect to risky web destinations, interact with unsanctioned AI services, handle sensitive files unsafely, or be manipulated through malicious prompt-based attacks. These risks are harder to manage when security teams lack consistent visibility and controls for agent traffic to internet, SaaS, and AI services. 

To give security teams a consistent way to inspect agent traffic at the network layer, in general availability today, Agent 365 extends Microsoft Entra network controls to Microsoft Copilot Studio agents and agents running on user endpoint devices, including local agents such as OpenClaw. These controls can help identify unsanctioned AI usage, restrict connections to only approved web destinations, filter risky file movement, and help block malicious prompt-based attacks before they lead to harmful actions. 

Confidently scale and govern AI agents while maintaining security and control 

Agent 365 extends even further beyond Microsoft platforms to discover, observe, govern, and secure local, SaaS, and cloud agents across your agentic AI ecosystem. Each of today’s announcements build upon Agent 365 capabilities we shared in March 2026 as well as detailed feedback of customers using the Frontier program, developers integrating with the platform, and partners testing Agent 365 capabilities. 

With Agent 365, we can scale and govern AI agents with confidence, while maintaining enterprise grade security and control. Agent 365 enables organizations to move beyond experimentation, driving tangible business value and innovation through trusted AI adoption. By providing a robust and integrated platform, Agent 365 empowers teams to confidently embrace AI and accelerate transformation across the enterprise.

—Yuji Shono, Head of the Global AI Office, NTT DATA Group Corporation, a global infrastructure, networking, and IT services provider.

As organizations begin to adopt Agent 365 at scale, we’ve collaborated with strategic partners to create targeted services to help customers onboard, tackle governance challenges and realize the platform’s full value.

Partner services offered today include expertise and guidance for: 

• Inventory and ownership: What agents exist, who owns them, and where they run.
• Least privilege: Right-sizing permissions and enforcing access guardrails without slowing delivery.
• Compliance and data protection: Preventing oversharing and producing audit-ready evidence.
• Threats and multi-platform estates: Understanding attack paths and governing across vendors and clouds.
• Ongoing operations: Lifecycle management, monitoring, and continuous governance hygiene. 

These valuable services are typically scoped as workshops and assessments (diagnose and roadmap), governance and enablement (stand up the control plane and guardrails), managed services (run and improve continuously), advisory and readiness (operating model and adoption readiness), and security and integration (harden posture and integrate third-party agents.)

• Find a Microsoft Agent Security and Governance partner to assist with your Agent 365 evaluation, deployment and adoption. Filter the directory by Offer: “Agent Security and Governance” and Country name. 

How to get started with Agent 365  

Agent 365 is now available in Microsoft 365 E7 or standalone at USD15 per user per month. Each Agent 365 license covers an individual who manages or sponsors agents, or uses agents to do work on their behalf, ensuring all agent activity is consistently governed across the organization in a way that’s predictable for scaled growth.  

In addition to the expertise of your Microsoft 365 team and partners, Agent 365 resources to support your experience include:

• The Agent 365 blog.
• The adoption hub and getting started guide.
• AI Skills Navigator for Agent 365. 
• Technical documentation on Microsoft Learn.

Plus, on Tuesday, May 12, 2026, a team of Agent 365 experts are hosting a live “Ask Microsoft Anything” to answer your questions about Agent 365—we hope you’ll join for the discussion.

Choose an ecosystem partner for agent security and governance

The post Microsoft Agent 365, now generally available, expands capabilities and integrations appeared first on Microsoft Security Blog.

We’ve shipped the new Run dialog – and we want to say thanks. As the team behind Windows Terminal and PowerToys, we’re excited to bring this to Windows 11, and your feedback helped shape it!

Run has been rebuilt from the ground up:

• Modern design: A refreshed look that matches Fluent Design and Windows 11, with dark mode support.
• Faster than before: Perf was top-of-mind when rewriting Run, and with a **94ms median time-to-show time it’s faster than ever before (more on how we got there below)!
• Quick user directory access: You can now type ~\ to jump to your user directory, then keep navigating just like you would from the command line.

The modern Run dialog is slowly rolling out in current Insider builds of Windows as an opt-in feature, ensuring we are collecting your feedback. To enable the new Run dialog, you’ll need to:

1. Be on Windows Insider Experimental Channel.
2. Enable to Settings -> System -> Advanced and toggle on the new experience with the “Run Dialog” option at the top of the screen.

How we’ve built it: Data-and-community driven

Windows Run, also known as the Run dialog, is a surface that has been around for over 30 years. It has become a heavily relied upon tool for developers and advanced users alike. Users have decades of muscle memory where they hit Win+R, navigate through their Run history, and hit Enter to quickly access various paths and tools. We all have our favorite tool we launch there as well. For us, some of our favorites are wt (Windows Terminal), mstsc (Remote Desktop) and winword (Microsoft Word). But it’s more than jUsT a TeXt BoX tHaT rUnS tHiNgS. The Run dialog can handle navigating both local and network file paths as well. And everything it does, it does fast. Win+R opens the run dialog seemingly instantly.

If we wanted to modernize the Run Dialog to fit the modern Windows 11 design style, we had to make sure it did everything just as well as before. We needed to maintain the same performance while also keeping the user interface minimal, just as Windows 95 intended.

Data driven engineering

When we set out on creating the new experience, we knew the existing dialog was fast. We also knew we needed to be sure we deeply understood how you all used it. Modernize, be opinionated, and evolve it. To help evolve, we added a measure briefly to the dialog to see what was being used and to measure time-to-show. This confirmed learned a few key things that helped the design process.

1. The existing dialog median time-to-show is 103ms
2. The browse button has very low usage. 0.0038% of users have clicked that button with a sample of 35 million.
3. Validated users do use the dialog to paste text from the clipboard, then copy it again without running anything to scrub text formatting.

Knowing exactly what was being used and how fast it was helped form a baseline that we could use to build the next iteration. From here, the rest of the story can continue.

Building the new Run

This rejuvenation started a few years ago and here are some of them mockups from our earliest designs:

Many of these prototypes were heavily inspired by PowerToys Run and a key theme of doing work quicker with your keyboard. In fact, PowerToys has been our testing ground for a lot of these ideas over the last couple years. With PowerToys, we could iterate quickly on experiments with direct community feedback and really polish these ideas.

These mockups ultimately inspired the creation of the Command Palette (“CmdPal”) which is in PowerToys today! CmdPal was a hackathon project trying to create the next iteration of PowerToys Run – one that could ultimately be a part of the OS itself. With CmdPal, we could iterate on building a native WinUI 3 application that’s just as fast as the run dialog. You read that right – the new Run dialog is a C#/WinUI 3 application. It’s compiled with dotnet AOT, so that it can bring all the blazing speed of native code, with the safety and modern language features of C#.

By our measurements, the modern run dialog has a median time-to-show of just 94ms. This was a huge team effort – we’ve collaborated tightly with partners across the platform to get these UI surfaces loading snappy. Improvements we’ve made to the platform don’t just make Run fast, but they help make the whole OS more efficient. Overtime, we expect these numbers to improve as well as there is still room for improvement.

From PowerToys to Windows: CmdPal powers the new Run experience

CmdPal is a lot more than just a run dialog, of course. But the same code that powers CmdPal in PowerToys has now become a part of the new Run Dialog. In fact, the run command provider in CmdPal is exactly the same code as the new Run Dialog! Anyone who’s ever contributed to CmdPal has helped build a part of Windows. Our team strongly believes in the power of open source. Our community is amazing, and we want to thank you all for helping make the new Run Dialog amazing!

New feature: ~\ for quick home directory access and icons in the list

Ever wanted a super quick way to be in your user directory? Type ~\ and you’ll be in your home directory and can freely navigate from there just like you can from the command line.

Feedback to improve

The most important thing to our team is getting feedback and constantly iterating and improving. This dialog may seem small, but for some folks it is a mission critical part of your workflow. We want to be sure it works for everyone. Please use the Feedback hub (Or Win+F since we are all keyboard-first users here ).

Mike, Kayla, Jessica, Niels, Chris, Dustin, Clint and many more. Terminal and PowerToys team

The post The new Run dialog: faster, cleaner, and more capable appeared first on Windows Command Line.

---