Browse All Posts (124)

Reynald Adolphe, Brigit Murtaugh and Lee Murray introduce the VS Code Agents window (Preview), a dedicated interface for running and steering multiple agent sessions across projects. They explain why it exists, walk through settings and theming, demo the workflow, and discuss extensibility and design goals.

Allison announces that GitHub Copilot individual (non-enterprise) plans can now access evaluation models, and that these models may be used when Copilot is set to auto model selection. The post also shows where to disable evaluation models in Copilot settings and links to the supported models documentation.

Kiefer Sheldon explains a Fabric-based pattern for showing real-time, in-report alerts inside Power BI without email. The approach treats notifications as data, writes them to a Fabric SQL database, mirrors them into a Lakehouse, and surfaces them through a semantic model so each report can display targeted alerts.

Zoey Li explains why Azure Container Registry’s global endpoint can feel opaque with Traffic Manager routing, and how new Regional Endpoints let teams pin pushes and pulls to a specific replica region while keeping the global endpoint for automatic failover.

John Savill recaps the major Microsoft AI updates from May 2026, covering new model availability in Azure AI Foundry, platform features like evaluation and networking, and notable changes across Copilot Studio and GitHub Copilot.

LZhang shares an end-to-end, field-tested setup for running Anthropic’s Claude Code through Microsoft Foundry in VS Code, including the exact settings.json shape, required Azure RBAC roles, tenant/login pitfalls, and a troubleshooting matrix for common errors like “baseURL and resource are mutually exclusive.”

Thomas Maurer talks with Kyle Ikeda about the Microsoft Agent Pre-Purchase Plan and how Agent Commit Units (ACUs) can be used across Azure AI Foundry, Copilot Studio, Fabric, and GitHub to make AI agent spend more predictable, with guidance on purchasing, tracking consumption, and budgeting responsibly.

This week's AI roundup focuses on what it takes to ship and operate agentic systems in real environments, from Microsoft Foundry updates (evaluation, model choice, and private networking) to clearer build-time vs run-time agent architectures. MCP kept gaining ground as the integration contract for tools, prompts, and "docs as context", with new Azure Functions prompt triggers and dedicated MCP servers for SRE workflows and Microsoft Learn grounding. On the GitHub Copilot side, enterprise rollouts got more practical with Claude Opus 4.8 GA, model targeting rules, stronger memory controls, and usage metrics that separate access from adoption. We wrap with IDE workflow changes that push plan-review-refine loops, plus security guidance that maps OWASP agentic risks to concrete governance tooling.

This week's DevOps roundup connects three threads that show up everywhere in modern delivery: supply chain risk, agent-driven automation, and platform guardrails that actually enforce policy. Microsoft flagged new npm install-time attack campaigns, a reminder that lifecycle hooks inherit your CI and workstation permissions unless you tighten token scope and credential exposure. On the automation side, guidance and tooling updates pushed agents toward production discipline (tool contracts, grounding, eval gates, and auditability), while GitHub and Azure shipped governance knobs like Code Quality enablement APIs, CodeQL improvements, hard budget limits for GHAS, and security baselines as code for Windows and Azure Arc.

This week's GitHub Copilot updates focused on making agentic work easier to manage at scale, from new model options and tighter enterprise controls to longer-running sessions you can supervise across devices. Claude Opus 4.8 reached general availability with a temporary premium request multiplier to plan around, while model rules add org-level targeting for phased rollouts. On the workflow side, VS Code continued building an agent-first experience (Agents window, remote sessions, and remote control GA), and MCP examples showed how tools, permissions, and doc grounding can make agents safer and more reliable. We also saw practical steps toward predictable behavior and measurable outcomes, with improved memory controls and new adoption cohorts in the Copilot metrics API to connect spend to real usage.

This week's ML roundup focuses on tightening the path from data to deployed models, with Microsoft Foundry expanding model options and leaning into trace-based evaluation that works across clouds. On the data side, Microsoft Fabric added features that reduce day-to-day pipeline overhead, including incremental Delta maintenance, CDC in Copy job, richer IoT streaming metadata, and new preview tooling for Excel ingestion and scheduled Spark pools. We also look at practical building blocks around ML work, from governed data exploration in Data Formulator to persistent agent memory with SQL, plus an infrastructure take on single-GPU training at the 100B+ scale and a simpler approach to Python data pipelines with dlt.

Shawn Henry shares a curated list of Microsoft Build 2026 sessions focused on Microsoft Agent Framework and Microsoft Foundry, spanning multi-agent patterns, production-scale agent lifecycles, governance and risk controls, and observability/evals across open standards and tools.

Authorised Territory demonstrates how to use the Plan agent in Visual Studio 2026 with GitHub Copilot, focusing on an AI-assisted workflow that starts from an existing solution and uses the agent to plan work before making code changes.

Andrea Griffiths and Nick Taylor introduce Pomerium on Open Source Friday, covering how the open source identity-aware proxy secures access to internal apps and services with authentication, authorization, and zero trust access patterns.

Rick Strahl explains why ASP.NET Core cookie-auth logins can “disappear” after IIS app pool recycles: the Data Protection key ring isn’t persisting, so previously issued auth cookies can’t be decrypted/validated. He shows how to fix it by enabling Load User Profile or by explicitly persisting keys to a known location.

kinfey lays out a practical two-layer architecture for building and operating AI agents using Microsoft Agent Framework and Microsoft Foundry, with GitHub Copilot acting as a “coding agent” guided by versioned SKILL files. The post uses the ZavaShop workshop to show tools, MCP, workflows, evals, and deployment/ops guardrails across Python and .NET.

Nick Brady’s May 2026 digest covers Microsoft Foundry updates across models, evaluations, networking, and SDKs—highlighting trace-based evaluation for agents on any cloud, new catalog models (Grok 4.3, DeepSeek V4), GPT-5 reinforcement fine-tuning at gated GA, Managed VNET GA, Foundry Local 1.1/1.2, and new skills/toolboxes support in azure-ai-projects.

Microsoft Defender Security Research Team reports on a dependency confusion campaign using 33 malicious npm packages that execute an obfuscated postinstall payload to fingerprint developer and build environments, and shares concrete mitigations plus Microsoft Defender detections and hunting queries to find related activity.

lily-ma announces a public preview update to the Azure Functions MCP extension: a new MCP prompt trigger that completes support for MCP’s core server primitives (tools, resources, and prompts). The post explains what MCP prompts are, shows Python examples, and links to quickstarts and docs.

Allison announces an update to the GitHub Copilot usage metrics REST API that adds AI adoption cohorts, helping enterprises and organizations understand not just who is using Copilot, but which Copilot surfaces they’re engaging with and how usage evolves over time.

GitHub shares a weekly “The Download” roundup covering Microsoft Build 2026 session picks, ecosystem news (Anthropic, Deno 2.8, Bun’s Rust move), Microsoft’s Azure Linux 4.0 release, and the general availability of GitHub Copilot Remote Control for taking local sessions on the go.

Reynald Adolphe and Vijay Upadya introduce Chronicle, an experimental VS Code feature that stores your GitHub Copilot Chat history locally in SQLite so you can query past sessions, generate standup-style summaries, and get feedback on how you prompt.

leereilly highlights four AI-focused open source projects featured in the Open Source Zone at Microsoft Build 2026, with a quick overview of what each project does and why developers might want to check them out in person or online.

stclarke introduces Data Formulator 0.7, an open-source, AI-powered workspace for enterprise data analytics that focuses on governed data connectivity, context-aware agents, and iterative visualization refinement across long-running analysis workflows.

mamathews rounds up May 2026 community picks from Microsoft MVPs and Fabric Super Users, spanning Power BI/VertiPaq performance, Fabric lakehouse design choices, Fabric Data Factory foundations with ADLS, NLP embeddings in Fabric, capacity cost optimization, Logic Apps-based capacity scheduling, and SharePoint list mirroring into Fabric.

Rob Lefferts summarizes why Microsoft Defender for Endpoint was named a Leader in the 2026 Gartner Magic Quadrant for Endpoint Protection, highlighting recent product capabilities like attack disruption with predictive shielding, custom telemetry collection for advanced hunting, simplified onboarding for Windows and Linux, and sovereign-ready data controls.

John Edward explains where Microsoft Copilot Studio sits in the Power Platform, and how it connects conversational AI to apps, workflows, data, and analytics through tools like Power Apps, Power Automate, Dataverse, and Power BI.

The Visual Studio Code Team shares what’s new in VS Code 1.123, with updates to the Agents experience and chat workflows, including handing off chat sessions to the Agents window, sending attachment-only chat requests, and improvements to Copilot Cloud task rendering.

Amar Badal announces that Anthropic’s Claude Opus 4.8 is available in Microsoft Foundry, positioning it for complex coding work, multi-step agentic workflows, and deep reasoning across long, document-heavy enterprise tasks.

Ayushman Bhattacharya explains how Pollinations.ai runs open source maintenance at scale, focusing on the practical systems they use to bring in contributors and keep work moving. The session covers contributor workflows, AI-assisted PR review, app submission pipelines, and how maintainers structure day-to-day work.

Liam Hampton demonstrates how to use third-party coding agents (Claude and Codex) inside Visual Studio Code via GitHub Copilot’s agent tooling, including when to choose local vs. remote execution and a short end-to-end demo workflow.

Jesse Houwing shows how to automate GitHub Copilot AI Credits budgeting by assigning per-user budgets based on Microsoft Entra ID group membership, using a GitHub Actions workflow and a PowerShell script that calls the GitHub enterprise billing API via the GitHub CLI.

John Savill shares a quick weekly Azure update covering what’s new across the Azure platform, aimed at keeping practitioners current on recent changes and announcements.

Hidde de Smet lays out a practical KPI scorecard for teams adopting AI coding agents under usage-based billing, using GitHub Copilot’s AI Credits model as the concrete example. It focuses on measuring speed, quality, reliability, and spend together, with a rollout plan and data sources you can wire into a weekly dashboard.

Fokko at Work demos selected new features in Visual Studio Code 1.122, focusing on GitHub Copilot updates like BYOK without GitHub login, the Agents window, and the built-in browser with device emulation, with notes on how enterprise policies and pricing plans can affect availability.

aakarshdhawan walks through building a React + TypeScript single-page app with Vite and deploying the compiled output to Microsoft Power Pages using the Power Platform CLI, including routing gotchas, activation steps, Dataverse Web API usage, and key security considerations around permissions and web roles.

paggarwal introduces Engineering Squad, an open-source multi-agent framework that turns plain-text requirements into user stories, technical design, production code, and automated tests using Azure OpenAI and Foundry Local, with a self-correcting review loop and traceable run artifacts.

yuvmaz breaks down the MegaTrain paper’s approach to training 100B+ parameter LLMs on a single GPU by treating GPU memory as a cache and streaming layers from host memory/NVMe. The post connects the technique to Azure NC-series VM choices, storage throughput, PCIe constraints, and cost/performance trade-offs.

Microsoft Defender Security Research Team reports on the “Mini Shai-Hulud” supply-chain campaign where typosquatted npm packages execute during install to steal cloud and CI/CD secrets. The post breaks down the two-stage payload, key indicators, and concrete hunting and mitigation steps for defenders.

Randy Pagels explains how to reduce repeated prompting by capturing team conventions in a copilot-instructions.md file so GitHub Copilot can generate code that matches your repo’s standards, architecture expectations, and preferred testing and design patterns.