The Microsoft Ignite Session Catalog is live. Explore the catalog and build your #MSIgnite agenda. See what’s next in AI and cloud, go deeper with the experts shaping the technology, and discover practical guidance you can put into action: msft.it/6017agYcV

Getting all setup for #MCP Community Connect - #Bengaluru globalai.community/e/bd1o37l…

🚀 From our VS Code Learn library: Extending Agents! In this E2E tutorial, we'll explore key aspects of AI Agents with Tools, Tool sets, Permissions and Sandboxing. We'll also look at MCP (Model Context Protocol) and how to use it to connect your agents to external data sources. Check it out ▶️ piped.video/SN3JTq3Q9-M

Anyone fancy a trip to SF

Today in Microsoft Foundry, we are advancing a model-agnostic agent platform, now including voice agents, that lets teams adopt better AI as it emerges: preserving the enterprise systems, knowledge, tools, and controls that make agents useful to customers. Learn more: msft.it/6017a93Mz

Developers are asking for better tools on GitHub for autonomous agentic coding. At the same time maintainers feel overwhelmed and want more control. Both are important! Here's me at AGNTCon Europe sharing 2 of the features I've been using daily for my own agentic workflows💜

Just put AKS Flex Nodes (preview) through their paces for an upcoming deep-dive blog. 🧪 Spun up a cluster, added a flexpool, scheduled a pod, watched it land. Scale-to-zero + on-demand VM capacity without the usual node pool rigidity. Full write-up soon. #AKS #Azure #Kubernetes

Last week, we hosted the first #GlobalAIConference of the season in Budapest... And what a way to start!!! 🚀 What stayed with me most was the feeling that something is building. This was a fresh and much-needed spark for the local tech community. 💝 #GlobalAICommunity #MVPBuzz

🤖 Agent-First Development From our VS Code Learn library, check out this in-depth tutorial. @madebygps demos how to work effectively with AI agents in VS Code, covering agent workflows, governance, debugging, execution environments, and hands-on application development. ▶️ piped.video/K040jc0PQWI

Marcel is talking to us about Microsoft Agent Framework Workflows: From Zero to Hero

Microsoft Agent Framework and GitHub Copilot nitter.net/i/broadcasts/1MJgNbqrr…

New episode of @AzureCommUG is tomrrow at 7:30pm UK time Come join us meetu.ps/e/Qgz9V/gb6Lv/i #Azure #GithubCopilot

🚀 Come and join us at the Nordic Integration Summit In about 2 weeks I will be heading Stockholm for two days of everything integration, cloud, and AI. I will be talking about Azure SRE Agent, exploring how agentic AI can help streamline operations and enhance platform reliability. Register today and secure your place:
 Looking forward to catching up with everyone there 🙌 #Integration #Cloud #AI #Azure #NordicIntegrationSummit #IntegrationCloudAI #Microsoft #AzureSREAgent #Azure #Cloud #Networking @msdev @Azure #MicrosoftTechCommunity

🗓️ Weekly recap (Sep 12-Sep 19) for Copilot CLI 🚀 2 releases 🛠️ 119 improvements 119 features & enhancements in this release Top features: • Vim mode is now available to everyone for modal editing in the composer, showing current mode as you type 🧑‍💻 • Added /settings options to opt-in to context management tools for agents and subagents • Added /config command to open a sidebar configuration screen in the CLI • Added /sandbox command for managing network host allow/deny rules without replacing configured upstream proxy • Session and memory import commands now support the semantic JSONL interchange format • Added copilot instruction list and copilot LSP list commands, replacing older plugin-kind flags • Added --json output to copilot plugin list and marketplace list/browse commands • Added enable and disable commands to copilot plugin, copilot MCP, and copilot skill, replacing separate enable/disable flags • GPT-6 Astra support has been added • Managed sandbox sessions can now be disabled for the rest of the session from an approved bypass prompt 🚫 Enhancements: • /vim command or editorMode setting enables Vim modal editing with mode shown inline • Added 'concise' transcriptView mode to group tool activity into expandable work summaries • Streamer mode masks internal model names on /model, footer and diagnostics without restarting model initialization 🎭 • --add-dir now rejects non-directory and inaccessible paths uniformly before session start • Terminal color themes are now respected even with incomplete palettes • End and Ctrl+E keys move cursor to true end of wrapped lines for precise editing • Modified key chords (Ctrl+X, Alt+X, capital X) no longer trigger /tasks letter shortcuts to reduce conflicts • Scheduled prompts now show errors when they fail at runtime for clearer diagnostics • CLI automatically retries image prompts without images when providers reject image-limit requests • Authenticated model lists refresh automatically after startup auth hydration • MCP servers no longer fail loading when connected to a running IDE, improving plugin stability • copilot init now removes .github directory on exit if no instructions file was written • MCP servers maintain correct workspace source labels after folder trust changes • /permissions picker marks Allow all when --allow-all-tools, --allow-all-paths, and --allow-all-urls are all set ⚙️ • Resumed sessions now show completed reasoning as Thought instead of Thinking… • copilot --help now correctly describes --no-* options and --no-auto-update behavior • Restore keyboard input and screen output correctly when returning from external editor on macOS and Linux • Model lists refresh after signing in, switching accounts, or signing out Bug fixes: • Fixed Windows one-command sandbox bypass to approve policy-blocked writes without disabling whole session 🛠️ • Fixed failed new-session handoffs so unused clients stop polling and current session data is preserved • Prevent terminal from leaving mouse tracking mode active after session or CLI exit • Fixed exported session --share=~/notes.md to write to home directory instead of creating a folder named '~' • Fixed in-process IDE bridge rejection by config validation that caused plugin reload failures and session restrictions • Fixed scheduled prompt runtime errors showing better diagnostics • Fixed thinking shape sent to Cla… (truncated in original data but assumed fixed issue) • MCP turns continue correctly even if tool list refresh fails after tool change • Fixed restored keyboard input and screen issues after returning from external editors on macOS and Linux Misc: • Updated CLI to keep --no-auto-login and --no-sandbox options supported but hidden in help as before • Improved CLI documentation for --no-auto-update indicating it runs bundled version • Improved help text for copilot --help on --no-* option descriptions • Added semantic JSONL interchange format support for session and memory import tooling github.com/github/copilot-cl… #GitHubCopilotCLI

Goodbye friends 🫡 Gemini 3.7 Flash GPT-5.5 GPT-5.4 GPT-5.4 mini GPT-5 mini Grok 4.5 I do have to say GPT-5.6 Luna Extra High destroys at great price.