What's New¶
Track the latest labs, content updates, and improvements to the AI Agents Learning Hub.
✨ This page is auto-generated from the git history on every deploy.
October 2026¶
🚀 New Labs Added¶
| Lab | Title | Level | Path |
|---|---|---|---|
| Lab 085 | Prompt and Context Engineering in GitHub Copilot | L100 | — |
| Lab 086 | Customize Copilot: Instructions, Prompt Files, Custom Agents, and Skills | L200 | — |
| Lab 087 | Copilot + MCP: GitHub, Playwright, and Your Own Server | L200 | — |
| Lab 088 | Copilot CLI: The Agent in Your Terminal | L200 | — |
| Lab 089 | Spec-Driven Development with Spec Kit | L200 | — |
| Lab 090 | Testing, Refactoring, and Modernizing Legacy Code with Copilot | L200 | — |
| Lab 091 | Copilot Cloud Agent and Code Review: From Issue to PR | L300 | — |
| Lab 092 | Secure Code with Copilot: Autofix, Secret Scanning, and Agent Risks | L300 | — |
| Lab 093 | Copilot SDK: Embed an Agent in Your Application | L300 | — |
| Lab 094 | Copilot Governance and FinOps: AI Controls, Policies, Budgets, and AI Credits | L300 | — |
| Lab 095 | Measuring Copilot Impact: Metrics API, DORA, and SPACE | L300 | — |
| Lab 096 | Capstone — Agentic DevOps: Spec → Plan → Agent → Review → Deploy | L400 | — |
| Lab 097 | Interactive: Tokens, Sampling, and Context Windows | L50 | — |
| Lab 098 | Context Engineering for Agents | L200 | — |
| Lab 099 | LLM Evaluations 101: Golden Sets, LLM-as-Judge, and CI Regression | L200 | — |
| Lab 100 | OWASP Top 10 for LLM and Agentic Applications: Threat Modeling | L200 | — |
| Lab 101 | Token Economics: Cost, Cache, Batch, and Routing | L200 | — |
| Lab 102 | Fine-Tuning vs RAG vs Prompting + LoRA on a Small Model | L300 | — |
| Lab 103 | Deep Dive: Train a Mini-GPT on CPU (Optional) | L400 | — |
| Lab 104 | MCP 2026-07-28: Stateless Servers, Discovery, and Multi-Round-Trip | L200 | — |
| Lab 105 | MCP Apps: Interactive UI Inside Agents | L300 | — |
| Lab 106 | Microsoft Agent Framework: Orchestration Patterns | L300 | — |
| Lab 107 | How GitHub Copilot Works: Context, Models, Agents, and Data Flow | L50 | — |
March 2026¶
🚀 New Labs Added¶
| Lab | Title | Level | Path |
|---|---|---|---|
| Lab 001 | What are AI Agents? | L50 | All paths |
| Lab 002 | AI Agent Landscape 2025 | L50 | All paths |
| Lab 003 | Choosing the Right Tool | L50 | All paths |
| Lab 004 | How LLMs Work: Tokens, Attention, Training, and Hallucination | L50 | — |
| Lab 005 | Prompting and Context Engineering | L50 | — |
| Lab 006 | What is RAG? | L50 | |
| Lab 007 | Embeddings and Local Similarity Search | L50 | — |
| Lab 008 | Responsible AI for Agent Builders | L50 | All paths |
| Lab 009 | Azure OpenAI Service Quickstart | L100 | |
| Lab 010 | GitHub Copilot — First Steps | L100 | |
| Lab 011 | Copilot Studio — First Agent | L100 | |
| Lab 012 | What is MCP? Anatomy of the Protocol | L100 | |
| Lab 013 | Free LLM Backends — Foundry Local and Ollama | L100 | — |
| Lab 014 | Hello Agent with Microsoft Agent Framework 1.0 | L100 | — |
| Lab 015 | Ollama — Run LLMs Locally for Free | L100 | All paths |
| Lab 016 | GitHub Copilot Agent Mode | L100 | |
| Lab 017 | Reliable Structured Output with Local Validation | L100 | — |
| Lab 018 | Tool Calling with Typed Local Contracts | L100 | — |
| Lab 019 | Streaming Responses: Events, Cancellation, and Safe UX | L100 | — |
| Lab 020 | Build an MCP Server in Python | L200 | |
| Lab 021 | Build an MCP Server in C# | L200 | |
| Lab 022 | Local RAG: Chunking → Embedding → Search → Answer | L200 | — |
| Lab 023 | Tools and Memory in Microsoft Agent Framework | L200 | — |
| Lab 024 | Teams AI Library Bot | L200 | |
| Lab 025 | VS Code Copilot Chat Participant | L200 | |
| Lab 026 | Agentic RAG Pattern | L200 | |
| Lab 027 | Agent Memory Patterns | L200 | |
| Lab 028 | Deploy MCP Server to Azure Container Apps | L300 | |
| Lab 029 | LangChain & LangGraph Basics | L200 | |
| Lab 030 | Microsoft Foundry Agent Service + MCP | L300 | |
| Lab 031 | pgvector Semantic Search on Azure | L300 | |
| Lab 032 | Row Level Security for Agents | L300 | |
| Lab 033 | Agent Observability with Application Insights | L300 | |
| Lab 034 | Microsoft Agent Framework Orchestration Patterns | L300 | — |
| Lab 035 | Agent Evaluation with Azure AI Eval SDK | L300 | |
| Lab 036 | Prompt Injection Defense & Agent Security | L300 | |
| Lab 037 | CI/CD for AI Agents with GitHub Actions | L300 | |
| Lab 038 | AI Cost Optimization | L300 | |
| Lab 039 | Vector Database Comparison | L300 | |
| Lab 040 | Production Multi-Agent Systems with Microsoft Agent Framework | L400 | — |
| Lab 041 | Custom GitHub Copilot Extension — Retired | ? | — |
| Lab 042 | Enterprise RAG with Evaluations and Foundry | L400 | — |
| Lab 043 | Multimodal Agents with GPT-4o Vision | L300 | |
| Lab 044 | Phi-4 + Ollama in Production | L400 | |
| Lab 045 | GitHub Copilot Workspace — Retired | ? | — |
| Lab 046 | Microsoft Agent 365 — Enterprise Agent Governance | L300 | |
| Lab 047 | Work IQ — Copilot Adoption Analytics | L200 | All paths |
| Lab 048 | Work IQ — Copilot Impact Analytics & Power BI | L300 | All paths |
| Lab 049 | Foundry IQ — Agent Tracing with OpenTelemetry | L300 | |
| Lab 050 | Multi-Agent Observability with GenAI Semantic Conventions | L400 | |
| Lab 051 | Fabric IQ — Real-Time Intelligence Agents | L300 | All paths |
| Lab 052 | Fabric IQ — Conversational Data Agent (NL → SQL) | L200 | — |
| Lab 053 | Fabric IQ — Batch AI Enrichment with AI Functions | L300 | All paths |
| Lab 054 | A2A Protocol — Build Interoperable Multi-Agent Systems | L200 | All paths |
| Lab 055 | A2A + MCP Full Stack — Agent Interoperability Capstone | L400 | |
| Lab 056 | Federated M365 Copilot Connectors with MCP | L300 | All paths |
| Lab 057 | Computer-Using Agents — Desktop Automation | L300 | |
| Lab 058 | Browser Automation Agents with OpenAI CUA | L300 | |
| Lab 059 | Voice Agents with GPT Realtime API | L200 | All paths |
| Lab 060 | Reasoning Models — Chain-of-Thought with o3 and DeepSeek R1 | L200 | All paths |
| Lab 061 | SLMs — Phi-4 Mini for Low-Cost Agent Skills | L200 | All paths |
| Lab 062 | On-Device Agents with Phi Silica — Windows AI APIs | L300 | All paths |
| Lab 063 | Agent Identity — Entra OBO Flow & Least Privilege | L300 | All paths |
| Lab 064 | Securing MCP at Scale with Azure API Management | L400 | |
| Lab 065 | Purview DSPM for AI — Govern Agent Data Flows | L300 | All paths |
| Lab 066 | Copilot Studio Enterprise Governance | L300 | All paths |
| Lab 067 | GraphRAG — Knowledge Graphs for Cross-Document Retrieval | L300 | All paths |
| Lab 068 | Hybrid Search — Vector + BM25 + Semantic Ranker | L200 | All paths |
| Lab 069 | Declarative Agents for Microsoft 365 Copilot | L100 | All paths |
| Lab 070 | Agent UX Patterns — Chat, Adaptive Cards & Proactive Notifications | L100 | All paths |
| Lab 071 | Context Caching — Cutting Costs for Large-Document Agents | L300 | All paths |
| Lab 072 | Structured Outputs Reliability Benchmark | L100 | All paths |
| Lab 073 | Agent Benchmarking with SWE-bench | L300 | All paths |
| Lab 074 | Foundry Agent Service — Production Operations Simulation | L300 | |
| Lab 075 | Power BI Copilot — Autonomous Analytics & Data Storytelling | L100 | All paths |
| Lab 076 | Microsoft Agent Framework — From SK to MAF | L200 | All paths |
| Lab 077 | AG-UI Protocol — Connect Agents to User Interfaces | L200 | All paths |
| Lab 078 | Foundry Local — Run AI Models Offline | L100 | All paths |
| Lab 079 | Deep Research Agents — Multi-Step Knowledge Synthesis | L300 | All paths |
| Lab 080 | MarkItDown + MCP — Document Ingestion for Agents | L200 | All paths |
| Lab 081 | Agentic Coding Tools — Claude Code vs Copilot CLI | L100 | All paths |
| Lab 082 | Agent Guardrails — NeMo & Azure Content Safety | L300 | All paths |
| Lab 083 | Multi-Modal RAG — Images, Tables & Charts in Documents | L300 | All paths |
| Lab 084 | Capstone — Build the Complete OutdoorGear Agent | L400 | All paths |
📦 Sample Datasets¶
All labs share a consistent OutdoorGear Inc. scenario with ready-to-use datasets:
| File | Contents |
|---|---|
data/products.csv |
25 outdoor gear products with categories, prices, specs |
data/knowledge-base.json |
42 RAG-ready documents: policies, FAQs, product guides |
data/orders.csv |
20 sample customer orders for RLS and order tracking labs |
🏗️ Infrastructure Templates¶
Deploy-to-Azure one-click buttons for three Bicep templates:
| Template | What it deploys |
|---|---|
infra/lab-028-mcp-container-apps/ |
Azure Container Apps + ACR for MCP servers |
infra/lab-030-foundry/ |
Azure AI Foundry + Storage + AI Services |
infra/lab-031-pgvector/ |
Azure PostgreSQL Flexible Server with pgvector |
🗺️ Roadmap¶
The hub currently tracks 107 lab entries across three languages. The 2026–2027 releases add:
- Copilot Free-first prompt and context engineering
- Repository instructions, prompt files, custom agents, and skills
- GitHub MCP Server, Playwright MCP, and Copilot CLI
- Spec-driven development with Spec Kit
- Free simulations for cloud agent, code review, governance, and FinOps
- An interactive token, sampling, and context-window lab
- A four-week GH-300 certification path
- Microsoft Agent Framework 1.0 and MCP 2026-07-28 modernization
- Labs 102–106, Challenge 012, AI-103/GH-600 maps, and the Instructor Hub
- Local/Real paths with cost and cleanup controls for simulated enterprise labs
See CONTRIBUTING.md or open a proposal.