Ashley Childress

A tutorial for gh stack, and for the fact that it does not care whether the diff gets read by my lead or by an AI. Small diffs get reviewed better either way.

Every AI redesign of my portfolio looked different and was the same filing cabinet underneath. What a style guide couldn't fix — and the one instruction that did.

AI coding agents leave you with 5-to-20-minute gaps all day. Five real ways I fill them — and the one "quick" habit that makes you the bottleneck.

I stopped treating permission prompts as the safety system, then spent four and a half months writing 134 standing rules to replace them. Nine things I do differently with AI, and the proof to back it up.

RAI footers record how much AI wrote every commit, but nothing reads them back. Here's the GitHub Action that scores nine months of git history.

A legacy Java service exhausted a finite identifier range, trapped HikariCP connections in an intentional loop, and blocked testing for weeks.

Snowflake fetches a repo's commit history, finds the one story in it with plain SQL, and narrates that single thread with Cortex. Six storylines. One card.

Spec-driven development hands AI the wrong job. The real win is arguing the design out in chat first, then letting the spec fall out of the conversation.

How I used Gemini in Save the Sun: it reads players' plain-language questions and plays a Norse wolf tuned to lose like a kid—never to cheat. DEV June Game Jam 2026 entry.

An ADHD brain tracker built on Gemma 4 (E4B + EmbeddingGemma 300M) that observes and never grades. Native audio, multi-lens extraction, on-device after one model download.

The bare-minimum AI coding workflow—pick the right model, plan first, write AGENTS.md for agents not humans, and cross-check across LLMs before you blame the tool.

Your electricity traces back to a specific mine, a specific operator, a specific county. Unearthed names all three—and what it cost the mountain.

Legacy Smelter is a thermal decommission platform where a Gemini-powered dragon analyzes your code and files very serious, very unhinged postmortems.

A backend engineer builds an immersive narrated web experience with Canvas, WebGL, layered audio, and her own Appalachian voiceover. No frameworks. 13 ADRs.

Building Supascribe: a human-in-the-loop AI workflow where ChatGPT writes structured artifacts to Supabase with review safeguards.

I stopped reviewing code and let Google Gemini build my UI. Then I audited the results to see what autonomy in AI development actually produces.

A DEV Community dashboard that surfaces posts needing human replies so members can quickly find where to engage.

Why GitHub Copilot skills sometimes help—and sometimes hurt. Learn how scoped context and SKILL.md actually change agent behavior.

How I built a portfolio as a search system using Algolia, prioritizing retrieval speed, consistency, and real-world constraints.

Building a conversational interface on Algolia where chat acts as navigation, grounded in fast, predictable retrieval.

My Portfolio Doesn’t Live on the Page 🚫📃 Ashley Childress...

A systems‑first portfolio built with Antigravity—focused on outcomes over code, intentional constraints, and the work that lives beyond the page.

Your AI assistant can help write the code, but your git history still needs a human signature. Here’s how RAI footers make AI attribution explicit.

Why AI orchestration, context limits, and trust matter more than speed—and what building for the next five to ten years actually looks like.

A battle-tested approach to GitHub Copilot instructions—repo rules, priority stacks, validation loops, and why AI needs boundaries to be useful.

When ADHD energy drops into “done mode,” it’s not burnout—it’s a reset. Here’s what the spark cycle really feels like.

GitHub Copilot’s wild late-October 2025 release: agents, smarter reviews, embeddings, metrics, and that one “helpful” test that nearly went to prod. Chaotic, hilarious, and absolutely worth the read.

Learn how to stretch every GitHub Copilot premium request, avoid quota meltdowns, and still get reliable AI output. Includes tested models, workflows, and prompt tips.

Your AI assistant might be the most productive co-author you’ve ever had, but your git history doesn’t know it. Here’s how to fix that.

You’ve built your first GitHub Copilot c̶h̶a̶t̶ ̶m̶o̶d̶e̶ agent—now it’s time to tame it. Here’s how to set up, test, and share custom modes like a professional who occasionally yells at their AI.