nl2sh
Local-first natural-language-to-shell experiment using llama.cpp and whatisit to generate shell command suggestions from plain language. Shows practical local LLM usage, command UX thinking, and small-tool building.
Compact notes on systems, tools, and operational patterns I actually use.
Light side: workplace systems, administration, user-facing operations, and keeping business services stable.
Products and environments I support in real operational contexts.
How I approach broken or unclear systems without turning the page into a command list.
Dark side: self-hosted systems, automation, monitoring, local LLM experiments, and infrastructure learning by actually running things.
Infrastructure I run and maintain privately.
Tools and habits that make the lab repeatable instead of random.
Small projects worth mentioning without turning the page into a portfolio wall.
Local-first natural-language-to-shell experiment using llama.cpp and whatisit to generate shell command suggestions from plain language. Shows practical local LLM usage, command UX thinking, and small-tool building.
Evidence first · real tools over buzzwords · small reversible changes · keep services boring · document decisions · validate live · human-reviewed AI assistance