#AI

The regex-based structural diff extractors landed and worked – then hit a ceiling regex was never going to clear. Here’s how coco’s diff pipeline actually moved onto tree-sitter: sharper structural context for commit messages, a layered regex fallback, lazy-loaded grammars – plus the bonus the same parser bought us, syntax-highlighted diffs in the Workstation.

A small filesystem tweak that turned my open source projects into a single workspace my agents can actually reason about. Nested git repos, a centralized orchestrator, and a documentation loop that finally stays in sync.

A client needed help managing a few VPSes, so I wrote a small bash script. Three rewrites and 21 releases later, it’s Strut: an open-source CLI for deploying Docker stacks anywhere there’s SSH, built agent-first for the way I actually code now.

After building Spec Shaver as a CLI for trimming oversized OpenAPI specs, I kept finding myself wanting a visual way to do the same thing. SpecForge is what came out of that itch: a web app for importing, curating, and exporting OpenAPI specs in the browser.

Hitting OpenAI’s 1MB / 30-operation cap on Custom GPT actions when your real API has 150+ endpoints is maddening. Spec Shaver is the CLI I built to reduce a bloated OpenAPI spec down to the operations that actually matter, while keeping the schema valid.

Coco is the AI-powered Git assistant I built because I was tired of writing terrible commit messages at 11pm. The story of how it grew from a one-off script into something I now use daily across every project.

Augments started as a way to give my own AI workflows a consistent set of tools across machines. This is the story of what it does, what I left out, and the cybernetic-enhancements framing I leaned into.

A bash one-liner that lists your locally downloaded Ollama models, plus the equivalent in Python, Node.js, Go, and Rust. The command I kept rewriting until I just dropped it into a shell alias.

Overcome information overload with AI tools like Fabric. Intro on how I’m starting to augment and enhance how I consume and interact with content on the daily to learn more in less time.

Every guide blames a fancy zsh prompt when your AI coding agent cannot read the terminal. So I deleted mine, then deleted bash too, and the commands still were not running. Here is where the trail actually led.

A Sunday backlog on coco that should have eaten the whole day took about an hour. Not because the AI wrote my code, but because steering files and specs meant it already knew the project instead of me re-explaining it every session. Here is what is in mine, and what I left out.

After years of building terminal apps and a growing inability to justify my GitKraken subscription, I built a Claude Skill for TUI design. Here’s what’s in it, why I made it, and what I’m planning to build with it next.

A weekend alone. Four projects. 222 commits in a single day. An honest look at what agentic coding is doing to my attention, what wisdom still costs, and how I am trying to stay balanced at light speed.

Wrapping QLC+ on a Raspberry Pi with the things that make a headless DMX rig pleasant to use: one-command provisioning, a friendly landing page, fixture groups, AI scene generation, and the boring reliability glue (udev rules, WiFi watchdog, doctor command) that turns lights-are-dead calls into a non-event.