I build AI-powered apps and automation workflows, and ship them with tests, CI, and live demos. 我開發 AI 應用程式與自動化工作流，每個專案都附帶測試、CI 與線上 Demo。
Currently looking for / 目前找尋：QA Automation · AI Application Testing · AI Workflow Automation · Junior AI Dev
unified-agent-spec-core — Strong-typed Governance CLI for AI Multi-Agent SOP
強型別 AI 多代理人 SOP 治理 CLI — 將 Markdown 交接編譯為機器可校驗的契約，內建確定性語意防偽 Oracle，在 CI 閘口硬性阻斷偽造的 Vitest 測試 log（
exit 1，不寫合規戳記）。終結 AI 代理人偷懶合謀。 Compiles Markdown handoffs into machine-checkable contracts; a deterministic semantic-forgery oracle hard-blocks fabricated Vitest logs at the CI gate.
Stack: TypeScript Node.js ≥20 Vitest JSON Schema GitHub Actions · agent-core init 一鍵注入治理鷹架（TASK/WORKLOG/JSON Schema/CI workflow）；agent-core check 在 PR 上硬性阻斷偽造的測試證據
WARDROBE AI — AI Outfit Recommendation PWA
Live Demo · API Health · DB Health
Upload clothes → AI recommends outfits by color season, occasion, silhouette, and material. 上傳衣物 → AI 依色季、場合、剪裁、材質推薦穿搭組合。
Stack: Expo Web Node.js Express Supabase/PostgreSQL JWT · Includes: Dockerfile, docker-compose, .env.example, production verification script
clothes AI — Daily Outfit Content Automation
AI generates outfit content → renders cards + model images → delivers via Telegram. Runs on a daily schedule. AI 每日自動生成穿搭內容 → 渲染卡片與模特圖 → 透過 Telegram 推送。
Stack: Python Playwright pytest Telegram Bot API · Real generated image outputs included in repo
ai_agent_sop_toolkit — AI Dev SOP Scaffolding CLI
pip install→ generates project rules, task tracking, agent roles, and handoff docs for Cursor / Claude Code projects. 安裝後執行ai-sop init，5 秒內在任何新專案建立完整 AI 協作 SOP 結構。
Stack: Python Click CLI · Bilingual docs (中文 / English)
TaskPilot Claude — AI Workspace for Claude CLI
Electron desktop app for managing Claude CLI sessions with persistent history and SOP-driven task execution. 桌面 AI 工作區，管理 Claude CLI 工作階段、持久化歷史記錄與 SOP 驅動的任務執行。
Stack: Electron React SQLite node-pty
微觀神聖領域：免疫風暴 — Roguelike 卡牌構築遊戲
非對稱雙陣營 Roguelike 卡牌構築遊戲。玩家指揮免疫系統對抗病原體，在 5×2 戰術棋盤上進行卡牌戰鬥。 Asymmetric Roguelike deck-builder — command the immune system against pathogens on a 5×2 formation board.
目前進度 / Progress： P4 Battle Presentation（4.7 / 4.9）— 戰鬥引擎 ✅ · 卡牌系統 ✅ · HUD ✅ · 陣型 UI ✅
Stack: Unity 2022.3 C# ScriptableObjects · Multi-agent Builder–Tester workflow · GitHub Actions CI
I use AI agents to accelerate development, but verify everything with tests, CI, production health checks, and deployment logs. Every project I ship includes a README strangers can follow, demo URLs, and reproducible setup.
我用 AI 助手加速開發，但每個環節都自己驗證：測試套件、CI、線上端點健康檢查、部署日誌。 每個作品都有完整 README、Demo 連結與可重現的環境設定，讓陌生人也能跑起來。