Ai on // TODO(craig): name this bloghttps://blog.beaumo.net/tags/ai/Recent content in Ai on // TODO(craig): name this blogHugoen-GBWed, 04 Mar 2026 00:00:00 +0000Give Your AI Assistant the Docs, Not the Search Barhttps://blog.beaumo.net/blog/local-docs-submodules-for-aiWed, 04 Mar 2026 00:00:00 +0000https://blog.beaumo.net/blog/local-docs-submodules-for-aiWeb search hands an AI assistant stale forum posts and deprecated APIs. Shallow submodules of the Home Assistant and HACS docs, symlinked under <code>docs/</code> and described in <code>CLAUDE.md</code>, give it the real thing.Learning in Reversehttps://blog.beaumo.net/blog/learning-in-reverseWed, 25 Feb 2026 12:00:00 +0000https://blog.beaumo.net/blog/learning-in-reverseAn AI assistant showed me C#&rsquo;s <code>$$</code> raw interpolated strings while I was getting something else done. Learning now happens sideways, and knowing a feature exists doesn&rsquo;t replace judgement about when to use it.Safer AI Scripting: Using Deno Permissions as a Trust Boundaryhttps://blog.beaumo.net/blog/safer-ai-scriptingSat, 10 Jan 2026 00:00:00 +0000https://blog.beaumo.net/blog/safer-ai-scriptingHave the AI write two files: a short wrapper you read and approve, and a Deno script that runs with only the permissions the wrapper grants. What that protects against, what it doesn&rsquo;t, and why <code>--allow-run</code> breaks it.The Negation Paradox: Why 'Don't Do X' Makes AI Do Xhttps://blog.beaumo.net/blog/negation-paradoxFri, 09 Jan 2026 01:00:00 +0000https://blog.beaumo.net/blog/negation-paradoxTelling a model not to do something puts that thing in its context, and attention has no way to subtract it. Describe the output you want, not the one you don&rsquo;t.Instructions for AI Coding Assistantshttps://blog.beaumo.net/blog/instructions-for-ai-coding-assistantsThu, 08 Jan 2026 21:00:00 +0000https://blog.beaumo.net/blog/instructions-for-ai-coding-assistantsA starting template for the instructions file an AI coding assistant reads before it touches your repository. It gets what you know about boundaries and shared code out of your head and into a file the tool reads.Teaching AI to Code Like a Senior Developer: SOLID Principles as Guardrailshttps://blog.beaumo.net/blog/solid-principles-as-ai-guardrailsThu, 08 Jan 2026 20:00:00 +0000https://blog.beaumo.net/blog/solid-principles-as-ai-guardrailsAI coding assistants are too helpful: the code works, but it doesn&rsquo;t belong. SOLID principles, used as constraints on the assistant, give it the pause an experienced developer has before changing things.