Build an AI Agent Platform in .NET 10 with GitHub Copilot
Welcome to OpenClawNet! This repository contains session materials and working code from the Microsoft Reactor live series on building production-ready AI agents using .NET 10, Blazor, and GitHub Copilot.
📌 This repo grows with each session. Code is added incrementally — after Session 1, you'll have a working chatbot. By Session 4, a full AI agent platform.
| Language | Series Page |
|---|---|
| 🇺🇸 English | Building an AI Agent Platform in .NET 10 with GitHub Copilot |
| 🇪🇸 Español | Construye una Plataforma de Agentes AI en .NET 10 con GitHub Copilot |
Architecture, Local LLM Providers, SignalR Streaming, Blazor UI
Walk through a pre-built .NET 10 solution: model abstractions, local LLM integration (Ollama/Foundry Local) with SSE streaming, EF Core storage, Gateway API, and a real-time Blazor chat UI — all orchestrated with Aspire.
- 📦 Materials available on session day
- 🇺🇸 Register · 🇪🇸 Registrarse
Tool Framework, Security Model, Agent Orchestrator Loop
Upgrade the chatbot to an agent: add a tool framework with approval policies, built-in tools (FileSystem, Shell, Web, Scheduler) with security gates, and the core agent reasoning loop.
- 📦 Materials available after Session 1
- 🇺🇸 Register · 🇪🇸 Registrarse
Markdown Skills, Context Management, Semantic Search
Give the agent personality with Markdown+YAML skills and long-term memory through conversation summarization and local embeddings for semantic search.
- 📦 Materials available after Session 2
- 🇺🇸 Register · 🇪🇸 Registrarse
Cloud Providers, Job Scheduling, Testing, Production Readiness
Connect Azure OpenAI and Foundry providers, add cron-based job scheduling, run the full test suite, and see the complete platform in action.
- 📦 Materials available after Session 3
- 🇺🇸 Register · 🇪🇸 Registrarse
# Clone the repo
git clone https://github.com/elbruno/openclawnet.git
cd openclawnetSession code and materials are published on the day of each live session. Register at the links above to be notified!
- .NET 10 SDK — Download
- VS Code + GitHub Copilot — Setup
- Ollama — Download or Foundry Local — Docs (run
ollama pull phi4-minior use Foundry Local SDK) - Docker — Docker Desktop
For Session 4 (optional): Azure subscription — Free account
Blazor Web UI ──SignalR──▶ Gateway API ──▶ Agent Orchestrator
│
┌──────────────────┼──────────────┐
▼ ▼ ▼
Model Provider Tool Framework Skills System
(Ollama/Foundry (File/Shell/ (Markdown +
Local/Azure) Web/Schedule) YAML)
│
▼
SQLite Storage
💡 Session 1 starts with just the chatbot path (UI → Gateway → Model → Storage). Each session adds a new layer until you have the full agent platform.
Each session uses an Explain → Explore → Extend approach:
- Explain — Walk through the pre-built code and architecture
- Explore — Run live demos, test endpoints, see behavior
- Extend — 2-3 small Copilot completions to add features
The code grows incrementally:
| After Session | What's in the repo |
|---|---|
| 1 | Chatbot: Models, Local LLMs, Storage, Gateway, Blazor, Aspire |
| 2 | + Agent: Tools framework, security gates, agent loop |
| 3 | + Personality: Skills system, memory, summarization |
| 4 | + Production: Cloud providers, scheduling, tests |
- Discord: Azure AI Community (.NET channel)
- Issues: Open an issue
- Resources: Generative AI for Beginners .NET · Aspire
MIT License — see LICENSE for details.
Built with ❤️ for the .NET and AI developer community. 🦀
