Hands-on prompt engineering and LLM evaluation. Built around a diagnose-then-ship loop: forensic transcript critique ▎ that names the exact failure mode, paired with a fixer that produces complete, deployment-ready rewrites — not ▎ vague advice.
Paste a diff and get a risk-ranked checklist of what it'll actually break — hidden dependencies, config drift, and blast radius that static analysis can't see, checked before you ship.
给非法律专业人士的「合同体检单」：输入合同文本，输出八段结构化清单——风险条款逐条原文摘抄 + 嫌疑说明，缺失项客观列举，让不懂法的人也能 3 分钟看清一份合同的关键信息与潜在风险。
解決 Electron 桌面程式在 Windows 上編譯失敗、打包出錯的疑難雜症，含免安裝exe發布教學。 如果你跟我一樣也打算自己開發桌面小程式，那麼這個妳一定要試，這是經由我本人踩過的坑的所有經驗，可以幫助妳省下不少時間跟TOKEN
For one person publishing their own content on a schedule: describe your setup and get an operations kit that makes silent failures visible — a status board, a runbook with idempotent retry and queue-fallback patterns, an incident ledger, and a daily health-check specification.
Most broken skills are not broken. They load, they validate, and they are never invoked - because the description does not match how anyone actually asks. Covers the one-character YAML mistake that makes frontmatter load empty with no error, how to write a description as a match target rather than a summary, and how to prove invocation in two steps instead of hoping.
A scan tells you what is wrong. This tells you what to do. Same engine as the ADA Website Risk Scanner — 14 WCAG 2.2 AA check families. But instead of a defect list, you get a work plan: findings grouped into batches, ordered by exposure reduced per hour of developer time, each with before/after code, effort estimate, required skill, and a verification step. Three phases. Quick wins first. A project manager can assign it as-is.
Turn any product or technology idea into a feasibility-checked prototype plan with architecture, BOM, budget, build stages, validation tests, risks, and an upgrade roadmap.
I'm a full-stack engineer who's shipped and maintained Next.js (App Router), React, and TypeScript apps under real production load. I kept watching AI write code that looked done - then leaked NEXT_PUBLIC_ secrets, swallowed errors, and skipped every state but "success." Shipwright distills the senior-engineer checklist I reach for when things break at 2 a.m. - secrets, auth, input validation, caching, and the unhappy paths - so your AI ships production-grade code, not demos.