Link Dump #242

If AI makes you more productive then you should have a bit more time for reading:

1. #BookOfTheMonth
   Software Architecture Metrics
   
2. Software Architecture
   1. The "Day 2" AI Problem: Why Standard API Gateways Fail at GenAI Scale
      While standard gateways are great at counting requests, they are completely blind to token consumption, streaming payloads, and prompt caching. Learn why moving from a GenAI prototype to enterprise scale requires moving past legacy HTTP proxies toward token-aware infrastructure before your next cloud bill arrives.
   2. Domain Types: What if you build nothing? #PickOfTheWeek
      Every architectural decision comes with a cost. Sometimes the most pragmatic solution is not another service, application, or vendor, but a simple manual process. 
      
3. Software Development
   1. bliki: Vibe Coding
      While pure "vibe coding" is a fantastic superpower for non-technical creators building disposable internal tools, professional systems require "Agentic Programming" - where the human acts as an aggressive code reviewer and structural gatekeeper. Find out where to draw the line in your AI stack.
   2. Dark Factories, Cobots, And The Potential Future Of The PR #PickOfTheWeek
      In the world where pull requests are created faster and are getting bigger, the author presents the model of cobot modules combined with dark factory modules as a one potential future that requires bringing together brand new and some very old ideas.
   3. Why DORA Metrics Look Different When AI Is Part of Your Development Workflow
      Your deployment frequency is up, your code-typing time is down, and your metrics look fantastic on paper. But author warns that AI-assisted workflows are creating a dangerous illusion of progress. By shifting bottlenecks to code review and introducing shared blind spots into automated test suites, AI coding tools are skewing traditional DORA metrics.
      
4. Testing
   1. An Expensive Errand: Chasing 100% Mutation Coverage with an Agent Can Make Your Test Suite Worse #PickOfTheWeek
      This deep dive into Codesai's refactoring experiment breaks down the critical distinction between relevant and irrelevant surviving mutants. Learn why treating the class as the absolute unit of testing leads to mock trauma, why agents struggle with context boundaries, and how to safely guide AI code validation without drowning in test bloat.
      
5. Clean Code
   1. What (almost) Everyone Gets Wrong About TDD & BDD #PickOfTheWeek 
      This article explores how the original concept of "customer-written check cases" from the 1950s evolved into the outer and inner loops of modern software engineering. Discover why separating specification from implementation is the only way to build a reliable "living contract" for your code.
   2. Help the Agent to Feel
      Human developers often rebel or quit when forced to adhere to hyper-opinionated coding standards, but an AI agent won't. The author explains how he uses a CLAUDE.md engineering manifesto alongside automated quality gates to align an LLM with his precise personal code architecture. Find out how establishing custom static checkers can transform a clumsy AI generator into a highly disciplined aesthetic programmer.
      
6. Leadership
   1. What We Tolerate
      Culture is driven entirely by concrete, observable actions. When a high-performer breaks an operational standard, your response sets the tone for the entire organization. Discover why your corporate values aren't written on your website - they are defined by the exact behaviors you choose to walk past.
   2. Why Intelligent Leaders Still Make Bad Decisions #PickOfTheWeek
      This deep dive into the psychology of executive failure examines how long-term validation breeds a dangerous overconfidence that suppresses organizational curiosity. Discover how to actively stress-test your operational models to ensure you aren't playing yesterday's strategic game against today's reality.