
Edit Feedback Triage|剪輯回饋衝突台
Edit Feedback Triage|剪輯回饋衝突台
功能 / What it does
把多人、不同版本的剪輯回饋整理成有來源的修改工單。每條原始意見保留一列,標示重複要求、候選衝突、舊版時間位置與待補資訊;明確決策另留紀錄,不因接受一項要求就自動否決另一項。另做 AI 語意檢查與可選詢問草稿,不自動發送。輸出 Markdown、CSV、JSON 與輸入雜湊,方便核對每輪資料。
Turn multi-person, multi-version video feedback into a source-linked work queue. Preserve one row per original comment, flag repeated requests, candidate conflicts, stale time positions and missing information. Keep explicit decisions separately; accepting one request does not silently reject another. A separate AI semantic review and optional question draft are never sent automatically. Outputs include Markdown, CSV, JSON and input hashes for identifying each round.
提供什麼 / What to provide
提供當前剪輯版本名稱、片長,以及已去識別化的回饋文字或 CSV/文字檔。若有版本標籤、時間範圍與原始意見,一起提供。無須自行編寫 JSON,相容 AI 工具依套件格式整理;缺漏維持可見,不臆測。
Provide the current cut label and duration, plus redacted review text or selected CSV/text files. Include version labels, time ranges and original comments where available. Your compatible AI client maps the input to the included contract; you do not need to write JSON. Missing details stay visible rather than being guessed.
合成範例 / Synthetic example
五則意見可分成一項可執行草稿、兩項互相衝突的標題顏色要求,以及兩項待確認版本或位置的意見。買家補上明確選擇與新舊時間映射後,可重跑而不刪除來源軌跡。這是合成示範,不是客戶成效。
Five notes become one actionable draft, two conflicting title-color requests and two notes needing version/location clarification. After an explicit choice and an old-to-new time mapping are supplied, rerun the queue without deleting the source trail. This is a synthetic example, not customer results.
交付與需求 / Delivery and requirements
US$9.99 一次買斷 Download 原始碼套件,不是 Capafy 代執行聊天或影片服務。需自備相容 AI 工具、模型與帳號;模型費用另計。離線核對程式需 Python 3.10+,只使用標準函式庫,不含第三方 Python 套件、發布者伺服器、內建 API 金鑰、遙測或網路呼叫。交給 AI 的意見資料仍受該供應商設定與條款約束。
US$9.99 one-time Download source package, not a hosted Capafy chat/video service. Bring your own compatible AI client, model and account; model fees are separate. The offline helper requires Python 3.10+ and only its standard library. No third-party Python packages, publisher server, built-in API key, telemetry or network calls. Comments supplied to an AI provider remain subject to that provider's settings and terms.
限制 / Limits
最多 200 則意見、5 MB 輸入 JSON、24 小時片長。程式只比對結構化值,無法驗證核准身分、看片、辨識所有語意衝突或自動對齊版本。不含審片平台 API/匯入器、影格或 drop-frame 換算、自動剪輯、傳訊或發布。「可執行草稿」不等於客戶核准,仍須人工或 AI 語意複核及自行檢查所有結果。不保證營收、利潤、節省工時或零錯誤。
Up to 200 comments, 5 MB input JSON and 24-hour cuts. The helper compares structured values; it cannot authenticate approvals, inspect video, detect every semantic conflict or automatically align versions. No proprietary review-platform API/importer, frame/drop-frame conversion, automatic editing, messaging or publishing. An actionable draft is not client approval. Human or AI semantic review and output checks remain required. No guarantee of sales, profit, saved time or error-free decisions.
已測試範圍 / Tested scope
已做本機 Linux/Python 3.12 測試與解壓套件執行;51 個單元測試本輪重跑通過,另有歷史 17 個邊界檢查與兩輪合成回饋案例。這些不代表真實買家、所有作業系統、所有 AI 工具自動辨識或 Capafy 執行認證。
Local Linux/Python 3.12 tests and extracted-package execution were performed. All 51 unit tests passed in the latest rerun; historical validation also included 17 boundary checks and two synthetic feedback rounds. These are not live-buyer tests or certification for every OS, automatic discovery in every AI client, or Capafy execution.
外部 API
自備相容 AI 工具/模型 | Buyer-selected compatible AI client/model


