Things I ship — products, the skills that build them, and the stack underneath.
What we ship.
The writing and review parts of how I ship.
The same Claude Code skills I use to draft these essays and pressure-test my own engineering plans. Each is a markdown prompt workflow; install by cloning into ~/.claude/skills/.
crafting-portfolio-essays
Writing essays in your voice — not the model's average. Asks for samples of how you actually write and talk; builds a tone-and-soul reference and a per-piece spec.
shipshape
Dual-level plan and code review for AI coding agents — staff-engineer then principal-engineer pass. Works with Claude Code, OpenCode, Codex, Gemini, Cursor.
triple-persona-review
Three readers who don't agree, dispatched in parallel, then consolidated into a prioritised editing report.
How the work actually gets done.
An AI-native operating model, made personal. Five tools, one operator. The customised GSD harness sits on top of Claude Code — phases, plans, executors, verifiers, all tuned to my own cadence.
Claude Code
Primary build surface. Every line, every commit, every diligence brief written via subagents.
GSD
Customised harness over Claude Code. Discuss → plan → execute → verify, with my own slash commands and review gates.
GitHub
Public OSS plus private Frollie monorepo. Branch protection, PR review automation, Actions for CI & nightly diligence runs.
Convex
Reactive backend for Frollie Pro. Real-time queries across operations, inventory, sales, financials, and accounting.
Vercel
Frollie Pro operator dashboard, per-brand consumer apps, lecture site, and this portfolio — all on one Vercel team.