同時開五個 AI 對話之後，你就失去對它們的掌握了——哪個還在跑、哪個十分鐘前就卡住、哪個早就結束但忘記關。 這支開一個本機儀錶板把 Claude Code 和 Codex 的對話並列顯示。在線狀態用實際存活的 process 驗證，不是只讀設定檔；超過十分鐘沒活動的標橘色。 還有一顆服務狀態燈，分辨「我的對話卡住」和「服務整個掛了」。
在伺服器上跑量化策略？兩次登入之間你其實是瞎的——今天有沒有觸發、還活著嗎、這週表現如何？這套工具把你的策略搬上 Discord：即時訊號告警、心跳在策略默默死掉的當下就 ping 你、每週績效報告（勝率／回撤／獲利因子／期望值）。底層還有 pipeline 健檢，讓當掉的 cron 沒辦法假裝成「這週剛好沒訊號」。它只轉發你系統已產生的訊號——不做交易決策、非投資建議。自帶 Discord webhook（敏感憑證）。
I've spent years in fitting bays reading Trackman data — this Skill works the way a $200/hour fitting reads: paste your Trackman, GCQuad, Mevo+, R10, or sim numbers and get the one dominant diagnosis, an honest technique-vs-equipment verdict, exact spec directions when equipment is truly the problem, and a practice prescription with a measurable checkpoint. It will tell you NOT to buy a new driver when the data says strike training, which is exactly why you can trust it when it says the opposite
漂亮的回測不等於真的有 edge。多數散戶「會賺」的策略其實是過擬合——那個 Sharpe 是掃了一堆參數挑出來最好的、是停損停利湊出來的假象、或只是運氣好的一段區間。 策略 Edge 驗證器用一整套統計檢定,幫你分辨「真訊號」還是「資料探勘出的雜訊」,全部只跑在你自己的報酬序列或權益曲線上: • 平穩 Bootstrap — 這條曲線跟零 edge 的雜訊分得出來嗎? • Deflated Sharpe — 扣掉「掃 N 組挑最佳」的選擇偏差後還站得住嗎? • PBO / CSCV — 回測期最強的參數,樣本外還是最強嗎? • Hansen SPA & Romano-Wolf StepM — 一籃子規則裡的贏家撐得過 data-snooping 嗎? • Bar-permutation MCPT — 利潤是訊號賺的,還是出場路徑湊的? • 樣本外分割 — 多少 edge 撐得過 holdout? 命令列 + JSON 輸出、自帶測試、標註學術出處。不需市場數據源、不需 API 金鑰、不需你的策略邏輯——你的數字全程不離開你的電腦。
Wraps each phase of a long AI task in a start-check / work / end-check bracket — and re-anchors the original instructions at every phase boundary so the agent cannot drift from its rules or skip verification as the session grows longer.
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.
Upload pg_stat_activity and pg_locks CSVs. Get blocker chains, long transactions, a 15-minute action plan, and a reviewed cancel-versus-terminate record.
A historical backtesting and validation-research protocol for algorithmic trading strategies: it treats your entry/exit/TP-SL/capital/leverage as fixed test inputs, stress-tests them with real costs and anti-overfit gates (plateau rule, walk-forward, random control, per-year robustness), and reports a research verdict (VALIDATION GATES PASSED / FAILED / INSUFFICIENT EVIDENCE). Analysis only: never deployment, sizing, leverage, or allocation advice.