https://arran4.github.io/blog/categories/artificial-intelligence/Recent content in Artificial Intelligence on Arrans Technical blogHugo -- gohugo.ioentechnicalblog123@arran4.com (Arran Ubels)technicalblog123@arran4.com (Arran Ubels)Arran Ubels. This work is licensed under a <a rel="license" href="http://creativecommons.org/licenses/by-sa/4.0/">Creative Commons Attribution-ShareAlike 4.0 International License</a>.Thu, 06 Aug 2026 10:42:00 +0000https://arran4.github.io/blog/post/2026/032-go-subcommand-infrastructure/Thu, 06 Aug 2026 10:42:00 +0000technicalblog123@arran4.com (Arran Ubels)https://arran4.github.io/blog/post/2026/032-go-subcommand-infrastructure/A Go CLI often begins with a small main.go, a few flags, and a switch statement. Then it grows. Before long, cmd/ contains argument parsing, configuration loading, filesystem access, database calls, output formatting, and the actual application logic. Adding a command means copying another block of flag handling. Testing a command means pretending to invoke the entire executable. Changing the CLI framework risks touching the whole program. [go-subcommand](https://github.com/arran4/go-subcommand) takes a different approach.https://arran4.github.io/blog/post/2026/030-using-gap-md-for-llms/Thu, 06 Aug 2026 09:57:57 +0000technicalblog123@arran4.com (Arran Ubels)https://arran4.github.io/blog/post/2026/030-using-gap-md-for-llms/Working with Large Language Models (LLMs) on multi-module projects or complex codebases can sometimes feel like a high-wire act. You give the LLM a large set of instructions, and while it might accomplish the immediate tasks, you often find yourself wondering: Did it actually have all the context it needed? Did it silently skip over something because a dependency wasn&rsquo;t ready? To prevent the LLM from trying to stubbornly forge ahead when it lacks prerequisites—or worse, hallucinating a solution—I&rsquo;ve developed a prompting pattern that I use regularly.