htmx
Recent essays from the htmx project tackle the shifting economics of software and new patterns for AI-integrated hypermedia apps.
Understanding code costs more now — a new essay argues that while AI tools have made writing code dramatically cheaper, the cost of understanding code has risen in proportion.1 As generated codebases grow larger and faster than ever, the ability to reason about what code does — not just produce it — becomes the scarcer and more valuable skill for developers.
Hypermedia-friendly MCP architecture explored — a detailed walkthrough describes building an app for Apple and Garmin watch workout programming using htmx alongside the Model Context Protocol. The author demonstrates how hypermedia principles can guide MCP app design, reducing user friction by letting the server drive application state rather than requiring complex client-side logic.2 The post walks through a real workout-scheduling tool, showing how structured interval training workflows map naturally onto a hypermedia-first architecture.