- Bring back ops pride (source) talks about operations and development teams as adversaries rather than collaborators trying to build applications that are useful and reliable. I don’t love this framing but it is how some teams are setup to be. How can you tell which one you have?
- On using simpler models for tasks via cli tooling vs api calls and mcp servers the latter of which sometimes require the thinking models, Simon Willison’s blog
- It can be eye opening to hop into a system and look around sometimes to see what’s happening … or runnning :) - Rachel by the bay
- Coordinating large groups of people in a company is hard. Some energy necessarily goes towards keeping everyone moving in the same direction. Lorin
- Why Kubernetes is retiring Ingress NGINX: Sounds like ingress nginx used heavily by k8s users isn’t being maintaned well enough to keep the platform secure and will be removed even when there isn’t a viable replacement for what it’s been doing in a couple months. This must not have been a fun choice for anyone involved to make.
- My AI Adoption Journey: Knowing what sorts of things agents can do well sounds important and trying to build verifiers so that an agent can check its work too. AGENTS.md is something that should probably be in any context you want an agent to work in.
- Claude code finds security bugs in open source software and Thomas Ptacek notes this is a perfect fit for llms - scanning reams of code for simple, well known patterns of badness.
