Systems | Development | Analytics | API | Testing

I just finished building Speedscale’s eBPF-based component to capture and analyze network traffic in a Kubernetes cluster, and it forced me to confront some uncomfortable truths about observability. While there were certainly some challenges along the way, particularly in dealing with Go applications, the approach was relatively straightforward.

It was a normal day in the land of eCommerce. Birds were singing, dashboards were loading, and merchants everywhere felt cautiously optimistic. Then the ground trembled. A tiny glitch. A flicker. A warning log no one read. And suddenly— BOOM! Shopify burst out of the digital ocean like a gigantic scaly beast that woke up on the wrong side of the server rack. Checkouts froze mid-purchase. Product pages stopped producting. Merchants stared blankly at blank screens. The Commerce Kaiju had arrived.

TL;DR: Claude Code writes features fast, but you need integration tests. With proxymock MCP, Claude Code can pull real production traffic and validate your changes automatically—one prompt, no manual setup. Watch it catch bugs before production.

Every year, KubeCon offers a candid look at where the cloud-native community stands — the tools gaining traction, the pain points teams share, and the big gaps still holding organizations back. After a week of deep conversations, session hopping, and talking to dozens of platform teams, one theme became impossible to ignore: Platform engineering still isn’t doing enough testing. And even more surprising: many teams don’t think testing is their responsibility.

Cursor is fantastic at cranking out code changes. I recently used it to splice a brand-new downstream API call into one of our Go microservices, and the diff looked great. The unit tests finished before I lifted my coffee mug, yet I still had zero certainty the change would survive contact with real traffic. That gap is all about integration tests, so I paired Cursor with proxymock and the outerspace-go demo service to prove the behavior end to end.

Hi everyone, my name is Bailey Ahrens, and I’m a marketing intern at Speedscale! I just returned a few days ago from KubeCon + CloudNativeCon North America 2025. While it was only my second trade show, it felt like a huge step forward in my confidence, skills, and future direction. My first conference (API World) was all about stepping outside my comfort zone. KubeCon was where I started leaning into that confidence and finding my place in the tech community.

At a previous company, we had over 100 microservices. I’d make what seemed like a simple change to one service and deploy it, only to discover it broke something completely unrelated. A change to the user service would break checkout. An update to notifications would break reporting. We spent more time fixing unexpected bugs than shipping features. The problem was our test scenarios were too simple.

We’ll always remember 2025 as the year AI code assistants went big. Copilot, Cursor, Claude, Windsurf, whatever. Developers went from mistrusting these tools, to being expected to turn over much of their coding labor to them. Even if, according to an extensive Stack Overflow survey, only 3 percent of professional developers say they ‘highly trust’ AI coding tools.