💡 Looking for an easier way to browse these labs? Check out our interactive web portal for a better reading experience: Microsoft Copilot Studio Labs Portal 🌐
| Title | Download | Overview |
|---|---|---|
| Create your own web-based AI assistant with Agent Builder | agent-builder-web | Create a Copilot agent in Microsoft 365 Copilot Chat with Agent Builder. The agent acts as a learning companion grounded in official Microsoft documentation to assist users with questions about Copilot agents. |
| Give your public website chatbot a brain and make it an agent | public-website-agent | Build and publish a Copilot Studio agent that delivers rich, contextual answers by combining your content with real-time data. Create a branded assistant that handles multi-part questions, reflects your organization's tone, and can be deployed to a public demo website. |
| Create a Monthly Business Review (MBR) Agent | mbr-prep-sharepoint-agent | Build a conversational MBR agent using Microsoft Copilot Studio and SharePoint. The agent streamlines MBR preparation by extracting key insights, surfacing trends, and answering questions directly from uploaded decks. |
| Create AI Solutions with Standard Orchestrator in Copilot Studio | standard-orchestrator | Create your first agent in Microsoft Copilot Studio. Learn the fundamentals including topic creation, message and question nodes, testing, and deployment. |
| Build Progressive AI Assistants with Agent Builder in Microsoft 365 | agent-builder-m365 | Master agent creation from basic web-grounded assistants to advanced SharePoint-integrated agents with code interpreter and image generation capabilities. |
| Proactive Contract Alerts in Teams with Azure AI Search and Speech-to-Text | contract-alerts-azure-ai | Configure Copilot Studio agents to proactively notify users when new files are added to Azure Blob Storage, connect Azure AI Search as a knowledge source, and integrate Azure Speech-to-Text services to transcribe audio files. |
| Copilot Studio Agent Academy - Recruit Level | Agent Academy | Comprehensive 13-lesson curriculum covering agents, LLMs, declarative and custom agents, Topics, Adaptive Cards, Agent Flows, and deployment to Teams and Microsoft 365 Copilot. |
| Build Intelligent Agents with Knowledge Sources, Tools, and Topics | core-concepts-agent-knowledge-tools | Learn how to create an AI agent from scratch in Copilot Studio, enhance it with knowledge sources, extend its capabilities with tools, and build custom topics for structured conversations. |
| Master Variables, Multi-Agent Architectures, and Channel Deployment | core-concepts-variables-agents-channels | Manage conversation state with variables, build specialized child agents for modular solutions, and deploy your agent across multiple channels to reach users where they work. |
| Monitor Performance and Evaluate Agent Quality | core-concepts-analytics-evaluations | Use analytics to measure agent performance, create evaluation test sets to systematically assess agent quality, and run evaluations to drive continuous improvement. |
| Set yourself up for success & discover ALM best practices | setup-for-success | Learn ALM best practices in Microsoft Copilot Studio. Structure your solutions, manage configurations, and enable Git-based source control to avoid common pitfalls like manual rework, overwritten changes, and fragile deployments. |
| Create your own SharePoint-based AI assistant with Agent Builder | agent-builder-sharepoint | Create a SharePoint-integrated AI assistant using Agent Builder in Microsoft 365. The agent can analyze Excel data from SharePoint, reference policy documents, and deliver insights through dynamic charts and images using code interpreter and image generation. |
| Create an 'Ask me anything' agent for your employees | ask-me-anything | Build a multi-agent IT Support solution using Microsoft Copilot Studio. Your primary agent intelligently routes queries to specialized sub-agents for ServiceNow knowledge and incidents, delivering fast, accurate answers across systems like SharePoint and ServiceNow. |
| Bring Your Own Model (BYOM) to Copilot Studio | mcs-byom | Integrate models from Azure AI Foundry into Copilot Studio agents using the Bring Your Own Model (BYOM) capability, enabling specialized AI functionality tailored to your business needs. |
| Expense Claims Agent with Approvals (Human-in-the-Loop) | human-in-the-loop | Build an end-to-end expense claims solution. An employee submits an expense report through a conversational agent, which triggers an agent flow to extract key details and evaluate the claim against policy using Multistage and AI approvals. |
| Using Multi-Agent in Copilot Studio | mcs-multi-agent | Learn how to leverage multi-agent configurations to build scalable, contextually-aware solutions. Understand when to use connected agents versus child agents, how to logically group tools and knowledge, and how to overcome limitations in tools and topics. |
| Using Component Collections in Copilot Studio | component-collections | Create, share, and manage reusable component collections to package topics and knowledge across multiple agents. Learn how to share components across agents with automatic sync and control access with primary agent settings. |
| Copilot Studio Tools | mcs-tools | Extend your agents with powerful tools including custom connectors, agent flows for deterministic business logic, MCP servers for live data access, custom prompts for structured responses, and Computer Using Agents (CUA) for legacy system automation. |
| Application Lifecycle Management (ALM) in Copilot Studio | mcs-ALM | Master the complete ALM lifecycle - from structuring solutions with custom publishers and configuring environment variables to deploying across environments with Power Platform pipelines and source control integration. |
| Build an autonomous agent that uses Computer Use to retrieve data from legacy systems | autonomous-cua | Build an autonomous agent that retrieves financial portfolio data from legacy systems without API access by simulating human interaction using the Computer use tool. Integrates Microsoft 365 Outlook to automate email-based data requests and responses. |
| Build an Autonomous Account News Assistant Agent | autonomous-account-news | Build an autonomous agent that scans Salesforce for high-value opportunities, searches the web for related industry news, uses Deep Reasoning to evaluate relevance, and compiles curated summaries into structured HTML reports sent via Outlook. |
| Create an autonomous support agent | autonomous-support-agent | Create an intelligent autonomous support agent that automatically responds to user emails, retrieves real-time ticket updates from ServiceNow, and answers questions using your knowledge base - all without human intervention. |
| Build a Data Fabric Agent | data-fabric-agent | Build intelligent agent-to-agent communication between Microsoft Fabric Data Agents and Copilot Studio to unlock powerful e-commerce analytics capabilities. |
| Measure Success & Analytics | measure-success | Design your agent to track conversation outcomes and collect user feedback on AI-generated responses. Gain meaningful analytics to identify which knowledge sources drive the highest satisfaction and understand patterns leading to abandoned or escalated conversations. |
| Source control and deploy your agents | pipelines-and-source-control | Deploy Copilot Studio agents with confidence using Power Platform pipelines. Automate and govern your deployment process across multiple environments with built-in validation, managed environments, and Git integration. |
| Copilot Studio Kit | copilot-studio-kit | Automate quality assurance for your Copilot Studio agents. Configure and run automated test scenarios, validate agent behavior using multiple test types, and gain actionable insights through analytics and governance dashboards. |
| MCP Qualify Lead | mcp-qualify-lead | Connect to Dynamics 365 Sales with Model Context Protocol to identify and prioritize high-value leads, enabling sellers to focus their time and effort where it counts most. |
| Build and Connect Your Guild's Custom MCP Server | guildhall-custom-mcp | Deploy a custom Model Context Protocol (MCP) server locally and connect it to a Copilot Studio agent that can query quests, hire parties, and manage the Guild treasury on your behalf. |
| Dataverse MCP Connector | dataverse-mcp-connector | Create and configure a Copilot Agent with Dataverse MCP Server integration to read and update account information from Dataverse using natural language queries. |
| Create an 'Ask me anything' agent (30 min version) | ask-me-anything-30-mins | A condensed version of the Ask Me Anything lab. Build a multi-agent IT Support solution with ALM best practices, ServiceNow integration, and deployment into Microsoft 365 Copilot. |
| Model Context Protocol (MCP) & Copilot Studio | MCS-MCP Lab | Learn how to deploy an MCP Server and integrate it with Microsoft Copilot Studio for streamlined deployment and management. |
This repository features a sophisticated lab navigation system:
- Journey-based Navigation: Labs are organized into learning paths (Quick Start, Business User, Developer, Autonomous AI)
- Bootcamp System: Special curated sequence with custom ordering (1a, 1b, 2, 3a, 3b, etc.) for intensive learning
- Smart Previous/Next: Context-aware navigation that respects your current journey or bootcamp progress
- Responsive Design: Optimized for both desktop and mobile learning experiences
- Mobile TOC: Floating table of contents button on screens ≤ 1366px for easy navigation
- Multi-Level Navigation: Hierarchical TOC supporting both main sections (h2) and subsections (h3)
Access the bootcamp through the top navigation or visit Labs Portal - Bootcamp.
Visual & Interactive:
- Multi-Theme System: Choose between Rich (colorful) or Minimal (clean) themes, each with light/dark modes
- Mermaid Diagrams: Interactive diagrams that adapt to your theme preference
- Responsive Breakpoints: Optimized for tablets (1366px) and mobile devices
- Smooth Animations: Polished transitions and hover effects throughout
Content Quality:
- Professional Layout: Clean typography with Source Sans Pro font family
- Code Highlighting: Syntax highlighting for multiple programming languages
- Progressive Navigation: Context-aware previous/next buttons guide your learning path
- Search & Filter: Quick access to specific labs by journey, bootcamp, or topic
This repository uses GitHub Actions to automatically convert lab markdown files into professional HTML and PDF formats using Pandoc 3.1.3:
- HTML versions: Web-ready with styling and navigation
- PDF versions: Print-ready with embedded images, Mermaid diagrams, and syntax highlighting
- Local generation: PDFs can also be generated locally using Docker (see LOCAL_PDF_GENERATION.md)
- Generated files: Available in the
dist/directory and deployed to GitHub Pages
Key PDF Features:
- ✅ GitHub-style callouts ([!NOTE], [!WARNING], etc.) rendered with proper styling
- ✅ Mermaid diagrams rendered to static SVG
- ✅ Embedded images (no external dependencies)
- ✅ Syntax-highlighted code blocks
- ✅ Professional layout with page numbers and headers
🎯 Local/CI Parity (CRITICAL):
- Local PDF generation MUST match GitHub Actions exactly to prevent deployment failures
- Always test locally before pushing if you modify lab content, diagrams, or PDF generation
- See LOCAL_PDF_GENERATION.md for complete testing guide
# Test PDF generation locally before pushing
.\scripts\Generate-Labs.ps1 -GeneratePDFs
# Quick test for a single lab
.\scripts\Generate-Labs.ps1 -SingleLabPDF "your-lab-name" -GeneratePDFs- 📖 Full documentation: See Contributing Guide for technical details
- 📝 Lab template: Use the Lab Template when creating new labs
- ✍️ Authoring guide: Follow the Lab Authoring Guide to ensure consistency across all labs
- 🔧 Architecture: Automated workflows, dependency management, and troubleshooting
- 🚀 Getting started: Just edit markdown files in
labs/- documentation generates automatically!
- 🚀 Quick Start: See Quick Start Guide for immediate development setup
- 🏗️ Development Guide: See Development Guide for comprehensive setup and workflow
- 📋 Architecture Decisions: See ADR for architectural decisions and context
For local development and testing:
# Start the development server
docker-compose up -d jekyll-dev
### 🚀 Quick Start (For Developers)
**⚠️ IMPORTANT**: After cloning, you must generate content files first:
```powershell
# REQUIRED: Generate Jekyll content from lab sources
pwsh -ExecutionPolicy Bypass -File scripts/Generate-Labs.ps1 -SkipPDFs
# Then start development server
docker-compose up -d
# Access the site at:
# http://localhost:4000/mcs-labs/Note: The index.md files are auto-generated and not stored in git. Always run the generation script after cloning!
Note: The /mcs-labs/ path is required for local testing due to Jekyll's baseurl configuration.
labs/ # ← Source lab content (edit these!)
dist/ # ← Generated HTML/PDF (auto-created)
.github/ # ← Workflows and technical documentation
package.json # ← Dependencies for document generation