|
Sr. Content Developer at Microsoft, working remotely in PA, TechBash conference organizer, former Microsoft MVP, Husband, Dad and Geek.
|
Shipping GenAI is easy. Shipping it without a surprise bill, latency spikes, and “why did it call the big model for that?” incidents is the hard part.
This article is a practical architecture pattern for cost control as a first-class system requirement — built around three levers:
It is a quiet Tuesday afternoon until the latency spikes begin. In the world of modern software engineering, we have moved far beyond the era of simple server monitoring. We no longer just “build and hope.” Instead, we strive for Continuous Reliability, a state where our systems are designed to be interrogated, understood, and improved in real time. This is the essence of Observability.
To truly master a running system, we must look through four distinct lenses, often called the pillars of telemetry. Each provides a different chapter of the story, and together, they offer a level of visibility that transforms production from a “black box” into an open book.
It is the age of AI – while AI is evolving many aspects of human interactions, the impact on software is particularly interesting. There is an opportunity for developers to infuse apps with solutions powered by generative AI and large/small language models – the end result should be smarter apps and better user experiences. Modern AI is also a huge opportunity to streamline and automate developer workflows for better productivity – AI can do much of the mundane code writing, freeing up developers to review code and guide architectural decisions.
As software developers increasingly leverage AI Agentic workflows to ship code, context is absolutely the key – specialized instructions or tools can provide grounding and help AI perform repeatable tasks. With Agentic AI, two concepts come up a lot: MCP Tools and Skills – they’re related, but they solve very different problems.
If you are a developer who is a foodie or loves to cook, a useful way to think about them is through food prep:
Let’s unpack that.
MCP Tools = Ingredients 
MCP (Model Context Protocol) tools expose specific contextual capabilities to an AI agent. Each tool does one thing and does it well.
Developers should think of MCP Tools as AI Agentic know-how to pull off specific tasks:
In food terms, these are raw ingredients:
By themselves, the raw ingredients aren’t very useful. You can eat flour, but you probably shouldn’t – in the hands of a skillful chef, the same flour can be turned into a yummy cake.
Key traits of MCP Tools:
MCP Tools answer: 
“What can the AI Agent do?”
Skills = Recipe Cards 
Skills are higher-level behaviors and instructional guardrails. Skills define how and when tools should be used to accomplish something meaningful.
Developers should think of Skills as AI Agentic instructions towards accomplishing specific software tasks:
In cooking terms, a Skill is a recipe card:
A recipe doesn’t grow tomatoes or mill wheat — it needs the raw ingredients, but orchestrates their use.
Key traits of Skills:
Skills answer: “How should the AI Agent solve this problem?”
The Agent = The Cook 
The AI Agent sits in the middle and coordinates what needs to be done to pull off a task.
A good cook doesn’t need to memorize the whole cookbook, just have recipe cards handy to know the steps need to followed to make a great meal. And the right tools absolutely help – fresh ingredients direct from source are the best.
Doom scrolling on social media might surface outlandish claims:
Serious software developers know the value of cooking right – eating well and healthy. And access to a personal chef is amazing – that’s what AI Agents bring to the table.
How to choose between MCP Tools or Skills: “Why not use both?”
The distinction between MCP Tool and Skill matters – this is important as developers expose context to AI Agents:
| Concern | MCP Tools | Skills |
|---|---|---|
| Abstraction level | Low | High |
| Reusability | Technical | Behavioral |
| Ownership | Platform / Framework | App / Domain |
| Stability | Long-lived | Evolves often |
| Dev mindset | “What APIs do we expose?” | “What workflows do we want?” |
While both are handy, there are consequences when the line between MCP Tools and Skills gets blurred:
Clean separation between MCP Tools and Skills provides composability without chaos.
Practical Rule of Thumb: If you’re wondering where something belongs, ask:
Back to food terms:
Now, let’s ask our AI chef to cook up something amazing for dinner. Cheers developers!
Ready to boost your productivity and simplify cross-platform .NET development? AI can help and Uno Platform MCP (Remote & App) Servers bring the context – AI is grounded in Docs/best practices and has eyes/hands to interact with the running app. Try it today – any OS, any IDE or with any AI Agent.
The post Introduction to Contextual AI: MCP Tools vs Skills appeared first on Uno Platform.
Microsoft has started testing a refreshed web experience for Xbox Cloud Gaming that looks even more console-like. Xbox Insiders can now try out the preview UI, which includes updated navigation features, plenty of new animations, and a refreshed design. It all feels like a teaser of what's to come for Xbox consoles in the years ahead.
The existing Xbox Cloud Gaming interface got its last major overhaul nearly two years ago, when Microsoft added the social features and UI you'd normally find in the Xbox dashboard. This new design makes that existing UI feel even closer to an Xbox console, thanks to a variety of new animations, a new library …
Read more of this story at Slashdot.
Log in
