The Aspire Demo We Didn't Get to Show Live: Watching an Agent Harness Work We ran out of time during Session 4 of MafClaw to show the Aspire dashboard demo. And this is one of those demos that really benefits from seeing it, not just hearing "yes, we have observability." So here is the walkthrough we didn't get to do live: take a small C# agent, let Microsoft Agent Framework's Harness call a tool, and follow that work through…

Hi! During the live Q&A in the second session of the From Model to Agent: The Agent Framework Harness, Live in C# series on September 10, 2026, someone in the audience asked a practical question: What happens if an agent requests approval, but the user does not answer? Can the approval expire or retry? That question is easy to overlook in a happy-path demo.

Hi! Ok so I spend the full morning with this, so let's better write it down for Future Bruno: The deployment was green. The hosted agent was active. The Microsoft Foundry Playground could reach it. There were no authentication failures, container crashes, or model errors. There was just one problem: when I asked the C# agent, "Who are you?", it answered:

Hi 👋 My favorite demo for this library is intentionally small: Raw ASR transcript Cleaned by s1-mini you don't have any any any any change at all? You don't have any change at all. That any any any any collapse is the interesting part. This is more than deleting a list of filler words. It is transcript normalization: turning the rough text emitted by speech-to-text into something a human would actually write.

⚠️ This blog post was created with the help of AI tools. Yes, I used a bit of magic from language models to organize my notes and automate the boring parts, but the geeky fun is 100% mine. Hi! The GitHub Copilot app can use models from your GitHub Copilot subscription, and it can also connect to models deployed in Microsoft Foundry.

Hi! So, you’ve probably used a “claw” already. One of those CLI agents that plans, runs tools, remembers what it did, and more. Cool. But have you ever stopped to think about what’s actually going on inside one? 🤔 Turns out, a claw is really just an agent harness: a loop around a model, wired up with tools, planning, memory, approvals and observability.

TL;DR KnownModels.MagenticBrain is now ready for local .NET usage with ElBruno.LocalLLMs. ONNX package is available at: Original model source: This support is designed to power end-to-end agentic app experiences such as: Why this model is important (and what it is designed for) MagenticBrain is not just another general-purpose chat model. It is designed for agent orchestration…

TL;DR Fara1.5-9B now works end-to-end in C# with ElBruno.LocalLLMs. The validated multimodal ONNX package is published at: You can run image+text inference locally with LocalVisionChatClient and KnownModels.Fara15_9B. This closes the loop for end-to-end local agent experiences in: When Microsoft Research introduced MagenticLite, MagenticBrain, and Fara1.5, it made a strong case for practical local agentic experiences:

⚠️ This blog post was created with the help of AI tools. The geeky fun and the 🤖 in C# are 100% mine. TL;DR Someone asked me to add Inkling — Thinking Machines' shiny new open-weights model — to ElBruno.LocalLLMs. I did the evaluation, and the answer is: not for now. 🔴 Inkling is a 975B-parameter, Mixture-of-Experts, multimodal (text + image + audio) model.

This blog post was created with the help of AI tools. I used them to organize ideas and write a first draft, but the workflow, the code, the testing, and the 🤖 enthusiasm are mine. Hi! I like reading AI news. I do not like opening seven tabs, comparing five versions of the same announcement, and wondering whether the most important developer update was hidden under an article about a robot making coffee.

Hi! If you have been playing with Foundry Local to run AI models on your own machine, you probably ran into the same thing I did: you never really know what is running. You load a model from the CLI. Your C# app loads another one through the SDK. A Python script spins up its own. A .NET Aspire sample launches a proxy.

Hi! Part 1 of the CopilotHarness series Big models are great for heavy thinking. But what about simple questions — "rename this variable", "write a short docstring", "what does this function do"? Those don't need GPT-5 across the internet. They can be answered instantly by a model running on your own machine, offline, for free. This post shows you how to wire GitHub Copilot in VS Code to local models using three minimal proxies — one for Ollama, one for Foundry Local, one for Azure OpenAI — and how to run all three together with a single command.

Hi! If you are building local AI apps in C#, you quickly hit a practical gap: Foundry Local SDK is great for local/on-device inference. Microsoft.Extensions.AI and Microsoft Agent Framework expect an IChatClient. There is no first-party Foundry Local -> IChatClient bridge yet. That is exactly why I created this library Nuget: Repo: Sample: ElBruno.MAF.FoundryLocal/src/ElBruno.MAF.FoundryLocal.Console at main · elbruno/ElBruno.MAF.FoundryLocal…

Hi! For a long time, many of us used GitHub Copilot as if it were unlimited magic: autocomplete, chat, agent mode, code review, increasingly powerful models, massive context, and long-running sessions that sometimes felt like a pair-programming marathon. And it worked. Well, mostly. Now, with usage-based billing and AI Credits, many developers are seeing something that used to be mostly invisible: every AI interaction has a cost.

Hi! Microsoft Build brought a wave of new AI model announcements, and one of the fun ones for image generation is MAI-Image-2.5. So, of course, I had to add support for it in my .NET image generation library and CLI tool 😅 The latest version of ElBruno.Text2Image now supports: MAI-Image-2.5 MAI-Image-2.5-Flash This means you can generate images using the new MAI models from a .NET app or directly from the terminal with the…

One of the most interesting announcements in the recent VS Code 1.122 release is that Bring Your Own Key (BYOK) models can now be used without signing in to GitHub. VS Code now supports BYOK scenarios, including chat, tools, and MCP integrations, without requiring GitHub authentication, making enterprise, offline, and air-gapped workflows much easier to support. I originally came across this update thanks to a tweet from Pierce Boggan:

Hi! What happens when a Python-based AI agent and a .NET-based AI agent need to work together? That is the idea behind MAF-A2A-NVIDIA-NemoAgents ( a reference app that shows how to combine NVIDIA NeMo Agent Toolkit, Microsoft Agent Framework, Agent-to-Agent communication, and Aspire into one multi-agent workflow. The repo describes itself as a production-ready sample that demonstrates NVIDIA NeMo Agent Toolkit + Microsoft Agent Framework with A2A communication, orchestrated with Aspire.

TL;DR This was the final ElBruno.NetAgent experiment: GitHub Copilot CLI + SQUAD using Azure OpenAI GPT-5.5 BYOK against the same app-building challenge I previously tried with CPU-only local models, GPU local models, and GPT-5-mini. The good news: GPT-5.5 was clearly better at staying inside phase boundaries, following safety rules, reducing broad stabilization loops, and working with strict quality gates. Once I switched to economy-mode prompts, the flow became much more disciplined: usually one implementation agent, one reviewer, short loops, and no wild repo-wide rewrites.

TL;DR This is part 3 of my GitHub Copilot CLI + BYOK experiment series. After testing local models on CPU and GPU, I moved the same experiment to Azure OpenAI using GPT-5-mini through BYOK. Same repo, same goal, same safety constraints — but this time with a cloud model that made SQUAD practical again. And that was the first big win.

Hi! Some of these started as small pet projects. IE: a quick helper for a demo, a tiny tool for a conference, a library to avoid repeating the same code again and again, or one of those “I’ll just build this in one evening” ideas that somehow becomes a real thing. And now, thanks to GitHub Copilot, many of these experiments are becoming open source, free NuGet packages that I hope are useful to everyone.

TL;DR After my first CPU-only experiment with GitHub Copilot CLI and local models, I wanted to try the same idea again, but this time with a GPU-powered setup. The first experiment proved that running GitHub Copilot CLI offline with local models was technically possible. This second experiment showed something more interesting: With the right model, GPU acceleration, small bounded phases, and strict human checkpoints, local agentic coding can become practical enough to move a real .NET project forward.

TL;DR I tested GitHub Copilot CLI running offline with local models on a CPU-only Microsoft Dev Box. The short version: Ollama worked for basic offline prompts, but struggled with long-context agentic workflows and tool execution on CPU-only hardware. LM Studio worked better as the local OpenAI-compatible provider. google/gemma-4-e2b successfully executed Copilot CLI PowerShell tools. 32K and 64K context were the practical sweet spot for guided development.

Hi! If you're running Ollama locally on Windows—whether you're tinkering with LLMs, building local AI demos, or just curious about the overhead of large models—you've probably wondered: Is it still running? How much CPU is it chewing? Did that model load? Welcome to ElBruno.OllamaMonitor. It's a no-frills system tray app that sits in your Windows notification area and tells you, at a glance, exactly what your local Ollama instance is doing.

Hi! dotnet tool install --global ElBruno.AspireMonitor aspiremon What it is AspireMonitor is a Windows tray app that puts your Aspire AppHost status one click away. No browser tabs. No switching context. Click the tray icon, see what's running, Start/Stop your app, or pin the resources you actually care about in a compact mini window. Why it exists If you use Aspire, you know the dashboard is useful—but you don't want a browser tab open all day.

Hi! usually hide my taskbar clock while recording videos, and after doing that dance one too many times, I had one goal: save a few clicks. Then yesterday, during a GitHub Copilot CLI session (watch it here), I decided to pick up the ClockTray app and turn it into a CLI tool. And... here we are. 🚀

Hi! Two weeks ago, I shipped t2i — a terminal-first CLI for text-to-image generation. Today I'm excited to announce two major additions that make t2i even more powerful: GPT-Image-1.5 and GPT-Image-2 support — Microsoft's DALL-E 3 and next-gen models via Azure OpenAI AI agent skill integration — Teach GitHub Copilot and Claude Code to use t2i automatically TL;DR New models:

Hi! I just shipped t2i, a terminal-first CLI tool for ElBruno.Text2Image. Generate images from your shell in two commands — no UI, no browser, just a simple cli interface to image generation from the cloud. This is the Lite edition (cloud-only, ~2.4 MB on NuGet) — perfect for CI/CD pipelines, deployment scripts, batch jobs, and developers who live in the terminal.

Hi! So, I've been away from any scenario that involves and needs data mapping, until I get to OpenClawNet. So I got to the point that I need to map my entities to DTOs? I read this a lot >> runtime reflection doesn't play nice with AOT and trimming. 😅 So I SQUAD myselft and built: 👉 ElBruno.AotMapper A Roslyn source-generator library that creates all your mapping code…

Hi! When Microsoft announced MAI-Image-2, I immediately thought: "I need to add this to ElBruno.Text2Image. Today." So I did. 😄 MAI-Image-2 is Microsoft's new image generation model on Microsoft Foundry — high-quality generation, a synchronous API (no polling!), a 32K character prompt limit, and flexible dimensions. And it's already supported in ElBruno.Text2Image with the same clean interface you already know.

The Microsoft Bing Search team just dropped something big: 👉 They’ve open-sourced a new set of high-quality embedding models, designed to be competitive with state-of-the-art approaches — and ready to run in modern AI workflows. And here’s the fun part 👇 You can already use these models from C#/.NET. 🤖 From Announcement to .NET in Minutes I integrated these new models into a C# library:

Hi! Earlier this year, Andrej Karpathy tweeted about using LLMs as "knowledge compilers" — a mind-bending idea: instead of asking an LLM questions, feed it raw data (papers, code, images) and let it automatically build a structured, navigable knowledge base. No RAG. No vector databases. Just pure understanding compiled into a graph. Then I saw @socialwithaayan showcase graphify — a Python tool that does exactly this for codebases.

Hi! You know that feeling when you're building an AI pipeline or a RAG workflow and you realize: "Wait… I need to turn all these PDFs, Word docs, HTML pages, and random files into something my LLM can actually eat"? 😅 Yeah, me too. That's exactly why I built: 👉 ElBruno.MarkItDotNet A .NET library that converts files to clean Markdown. Think of it as the 

Hi! So Google just dropped Gemma 4 — their most capable open model family yet — and I couldn't resist. I spent a good chunk of time digging into the architecture, trying to convert models, hitting walls, finding workarounds, and hitting more walls. Here's where things stand with ElBruno.LocalLLMs. Spoiler: the library is ready for Gemma 4. The ONNX runtime...

Hi! I needed a quick way to generate QR codes in .NET—no heavy dependencies, no complex setup, just something that works and that’s exactly why I built: 👉 ElBruno.QRCodeGenerator An open-source, lightweight library to generate QR codes in multiple formats: 🖥️ ASCII (perfect for CLI tools) 🖼️ Images (PNG/SVG) 🔗 Payloads (WiFi, URL, email, vCard) 🎬 Watch the video…

Hi! Today I want to share something that's been hearing a couple of times: You know when you're building an AI agent or working with LLMs, and you have dozens (or hundreds) of tools available? What do you do? Send ALL of them to the LLM every single time, right? Yeah, me too. And it's expensive. I had as a pet project for a while, so I decided to finish it, and start to give a real try.

Hi! Are you ready to dive into the world of AI agents? Join us for an exciting four-session live series at Microsoft Reactor where we'll build OpenClawNet, a production-grade AI agent platform from scratch using .NET 10 and GitHub Copilot. What You'll Learn Over four interactive sessions, we'll transform a simple chatbot into an intelligent, autonomous agent capable of reasoning through complex tasks, calling tools, remembering context, and scaling to the cloud.

Hi! Let's look at this code snippet: using ElBruno.LocalLLMs; using Microsoft.Extensions.AI; using var client = await LocalChatClient.CreateAsync(); var response = await client.GetResponseAsync([ new(ChatRole.User, "What is the capital of France?") ]); Console.WriteLine(response.Text); That's it. This runs a local LLM. No API keys. No REST calls. The model downloads automatically the first time. Let me show you more. ⬇️ Download Progress and Model Info…

Una nueva tanda de episodios de No Tiene Nombre trae noticias sobre inteligencia artificial, tecnología y el impacto real que estas herramientas están teniendo en el mundo del desarrollo y mas. Aquí tienes un resumen rápido de cada episodio con su enlace directo. 🎧 NTN 487 Título: NTN 487 – IA y nuevas batallas tecnológicasDescripción: Reflexión sobre cómo los nuevos modelos y herramientas de IA están cambiando el equilibrio entre empresas tecnológicas, desarrolladores y usuarios.

Hi 👋 If you’re experimenting with AI Agents in .NET, one of the concepts that appears quickly is Agent Skills. Skills are the mechanism that allows an agent to extend its capabilities: calling prompts, running code, or executing external logic. To make this easier to understand, we published a small sample repo: 👉 The goal of the sample is simple: …

Sometimes you want to see the clock all the time.Sometimes you want it completely hidden. Depending on how you work, the Windows clock in the taskbar can either be useful… or distracting. So I built a tiny utility that lets you toggle the Windows clock visibility with a single click. Meet ClockTray. The Problem Windows lets you show or hide the taskbar clock, but the process is not exactly a quick one.

Hi 👋 What if you could build a real-time voice conversation app in .NET — speech-to-text, text-to-speech, voice activity detection, and LLM responses — all running locally on your machine? That's exactly what ElBruno.Realtime does. 🎥 Watch the full video here (coming soon) Why I Built This I've been building local AI tools for .NET for a while — …

Hi 👋 These days Microsoft announced FLUX.2 Flex on Microsoft Foundry, I immediately thought: "I need to wrap this for .NET developers." So I setup a SQUAD team and I did it. And then I thought: "Wait — I have a couple of Test-to-Image local pet projects, what if my SQUAD also help to polish and publish this? Same interface, and of course let's make it Microsoft Extensions for AI compatible"

Hi! Let’s look at these 2 code snippets… what’s behind them? 🧠 Snippet 1 — VibeVoice (Native TTS in .NET) using ElBruno.VibeVoiceTTS; using var tts = new VibeVoiceSynthesizer(); await tts.EnsureModelAvailableAsync(); // auto-download model if needed float[] audio = await tts.GenerateAudioAsync("Hello! Welcome to VibeVoiceTTS.", "Carter"); tts.SaveWav("output.wav", audio); This generates a WAV file from text using the VibeVoice-Realtime-0.5B model, running locally via ONNX.

Hi! Big milestone these days: The Microsoft Agent Framework (MAF) just reached Release Candidate status 🎉 Official announcement here:👉 As someone who has been building apps, samples, demos, orchestration experiments and livestream content around MAF for months… this one feels GOOD. Let’s talk about this. 🤖 What is Microsoft Agent Framework? The Microsoft Agent Framework is a .NET-first (and Python) framework to:

Hi! This post walks through a tiny but powerful scenario: ingest images locally, search them with natural language, and let an agent decide when to call tools. Everything runs on your machine, and it's an simple example of how to use Microsoft Agent Framework with local embeddings. We’ll use: CLIP image embeddings (ONNX) for fast local vector search. Ollama…

Hi 👋 If you’ve used ElBruno.LocalEmbeddings for text embeddings, you’re going to love the new image capabilities. I asked several friends about this, and they challenge me to give it a try, so here it is: ElBruno.LocalEmbeddings.ImageEmbeddings a library brings CLIP-based multimodal embeddings to .NET — fully local. It is powered by ONNX Runtime, and ready for image search and image RAG workflows.

Hi! 👋 One of the questions I get most often is: "Bruno, can I build a RAG (Retrieval-Augmented Generation) app in .NET without sending my data to the cloud?" The answer is a resounding YES. 🚀 In this post, I'll walk you through three different ways to build RAG applications using ElBruno.LocalEmbeddings — a .NET library that generates text embeddings 

Hi! If you’re running local LLMs using: Foundry Local Ollama Models like Phi, Qwen, or Llama At some point you’ll want to experiment with RAG. And that means one thing: 👉 You need embeddings. During production, you’ll probably rely on managed services and vector databases.But during experimentation? You just want something simple. That’s why I created:

GitHub Copilot just crossed a very interesting line. It’s no longer “just” helping you write code — it can now run as an agent, with goals, tools, and autonomy, using Microsoft Agent Framework (MAF). 🎥 Watch the full video here: In the video, I walk through three simple C# samples showing how Copilot can be used as an agent…

Persisting Microsoft Agent Framework Sessions with ASP.NET, Redis & Blazor Have you ever built a chat app where the AI forgets what you said five seconds ago? 😅That's the classic stateless API problem — and today we're fixing that in .NET using the Microsoft Agent Framework (MAF), persistent agent sessions, and a Blazor frontend so you can…