This week covers Microsoft’s open-source Agent Framework for agentic AI, prompt-injection risks and mitigations, and the causes of language model hallucinations. It also highlights NuGet package security updates, Azure SQL Data API Builder improvements, Reka’s new Parallel Thinking feature, and the latest in AI benchmarking.
Cloud
- Announcing Aspire 9.5 - .NET Blog (Jeffrey Fritz) - I'm genuinely excited to try it. Aspire is such a great tool 🔥.
Programming
- Publishing NuGet packages from GitHub actions the easy way with Trusted Publishing (Andrew Lock) - Nice post that[explains the details of this new secured process on Nuget (.NET package library)
AI
MCP Prompt-Injection: The Trust Paradox in AI (Saurabh Davala, Sundeep Gottipati) - This is a good post to learn to start learning about the real danger. To be aware of the potential risk and learning about what we can do.
Why language models hallucinate - Very interesting post that explains the reason why we still have hallucinations and how it works
Introducing Parallel Thinking for Reka Research (reka) - New feature released that improves the performance of our web research, fantastic!
Measuring the performance of our models on real-world tasks - Interesting idea of benchmark looking forward to see how evolves
Introducing Microsoft Agent Framework: The Open-Source Engine for Agentic AI Apps (Takuto , Shawn , Elijah) - Nice evolution of semantic kernel and AutoGen. It comes hot supporting .NET and Python.
Databases
- Data API builder 1.6: Advanced Behaviors with Special HTTP Headers (Jerry Nixon) - Nice quick post about more advanced features on the data API builder (aka DAB)
Miscellaneous
- We all have a choice (Salma Alam-Maylor) - It's all in the title
Sharing my Reading Notes is a habit I started a long time ago, where I share a list of all the articles, blog posts, and books that catch my interest during the week.
If you have interesting content, share it!
~frank