MAF AzureBrasil Hub

Hub com o código de cada vídeo da playlist do Microsoft Agent Framework no canal AzureBrasil.Cloud.

Repositório do hub ↗

Ep. 02

Sessões: salvar e retomar a conversa

Serializa a sessão do agente e retoma depois; versão com Persistent Agents do Foundry em src/MAF.02.AzureAiAgentPersistent.Console.

Ver código no maf-playlist ↗
Ep. 03

Agente numa API com tool e aprovação humana

Expõe o agente por HTTP com uma tool que exige aprovação (ApprovalRequiredAIFunction).

Ver código no maf-playlist ↗
Ep. 08

Microsoft Agent Framework - DevUI Demo

This repository contains the source code for the eighth video in the Microsoft Agent Framework playlist on the AzureBrasil.Cloud channel.

Ver repositório original ↗
Ep. 09

MAF Video 09 - Microsoft Agent Framework with Model Context Protocol

This repository contains the code for the ninth video in the Microsoft Agent Framework (MAF) playlist on the AzureBrasil.Cloud channel.

Ver repositório original ↗
Ep. 10

Ep. 10 — Long-term memory with ChatHistoryMemoryProvider

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 11

Agent Skills — Microsoft.Agents.AI

Example of using FileAgentSkillsProvider from the Microsoft.Agents.AI package to give an AI agent "skills" via Markdown files.

Ver repositório original ↗
Ep. 12

Ep. 12 — Reasoning models + tools (multi-hop)

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 13

Ep. 13 — Observabilidade com OpenTelemetry

Parte da série Microsoft Agent Framework · índice de todos os episódios no maf-azurebrasil-hub.

Ver repositório original ↗
Ep. 14

🧠 Compaction Pipeline — Conceitos (Agent Framework)

O Compaction Pipeline é uma estratégia para reduzir o histórico de conversas enviado para o LLM, mantendo o contexto essencial enquanto controla:

Ver repositório original ↗
Ep. 17

Mixed Workflow — Agents + Executors

Example of mixing AI agents and custom executors in a single workflow using the Microsoft.Agents.AI.Workflows package.

Ver repositório original ↗
Ep. 18

Ep. 18 — Workflow checkpoints: Resume vs. Rehydrate

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 19

Ep. 19 — Code Interpreter with Microsoft Foundry

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 20

maf-video-20 — Microsoft Foundry Memory

Demonstrates how an AI agent can learn a user's writing style from example LinkedIn posts and reproduce it faithfully — even across completely new sessions — using Microsoft Foundry Memory.

Ver repositório original ↗
Ep. 21

maf-video-21 — Vibe Coding Research Agent

A console-based AI agent built on Microsoft Agents AI and Azure AI Foundry that researches AI developer tools and the vibe coding ecosystem. The agent streams responses, supports plan/execute modes, persists file memory across sessions, and can fetch and convert web pages to Markdown on demand.

Ver repositório original ↗
Ep. 22

ToolbBoxTools — Code Review Agent with Azure AI Foundry Toolbox via MCP

A .NET 10 console application that demonstrates how to build an AI code review agent using Azure AI Foundry, Microsoft Agents SDK, and Model Context Protocol (MCP). The agent connects to a Foundry Toolbox backed by the Roslyn Analyzers repository (via GitMCP), then generates a structured markdown report for a given C# file.

Ver repositório original ↗
Ep. 23

Ep. 23 — Harness Agent with background sub-agents

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 25

Permission-Aware RAG with SQL Server VECTOR + Microsoft Agent Framework

This demo shows how to build a RAG agent whose retrieval is filtered by per-user file permissions. Each user only receives answers grounded in documents they have access to — enforced at the SQL query level using the native VECTOR type in SQL Server 2025, with an optional in-memory caching layer that can be invalidated on demand.

Ver repositório original ↗
Ep. 27

Ep. 27 — Foundry Hosted Agents (.NET)

> Part of the Microsoft Agent Framework video series on AzureBrasil.Cloud (PT-BR). All episodes: maf-azurebrasil-hub

Ver repositório original ↗
Ep. 28

maf-video-28

Sample code for a video walkthrough on evaluating AI agents built with the Microsoft Agent Framework and Azure AI Foundry.

Ver repositório original ↗