repowise-dev/repowise：README 來源編輯指南

repowise-dev/repowise 的 README 將專案描述為「Codebase intelligence for AI and humans: code health scores, auto-generated docs, git analytics, dead code detection, and architectural decisions via MCP.」。本文只整理倉庫可直接核對的內容，不把 star、Fork 或宣傳語當成品質證明。README 在「up to −96% tokens to load context  ·  −89% file reads  ·  −70% tool calls」下寫到：Free and self-hosted, runs on your machine, and the first index needs no API key.。這說明的是專案邊界，不是已完成的生產驗證。

從 README 的「Know what's dangerous before you merge」與相關條目，可以先判斷它是否處理你的實際問題：Bug history. Which files and symbols actually get bug-fixed, and how recently.。若需求不同，不應只因專案熱度就採用。本文保留原始專案名、命令與元件名，方便回到一手來源核對。 README 另外列出一項可核對的資訊：Change risk. Score any commit or base..HEAD range 0-10 from the shape of。這類原文條目可用來設計試跑步驟，但不能取代實際環境測試。

README 將運作方式分散在「up to −96% tokens to load context  ·  −89% file reads  ·  −70% tool calls」等段落。可確認的線索包括：repowise computes those answers once and keeps them current on every commit. Your agent reads the answer instead of the codebase, and the same index gives your team a defect-validated health score, change-risk scoring on every PR, and a。本文不把未寫出的架構、效能或安全邊界補成結論；真正的執行鏈仍要配合目錄、設定檔與版本標籤檢查。

第一次安裝應從 README 指出的入口開始。目前可核對的命令是： repowise distill pytest # 61% fewer tokens, all 11 failure lines kept repowise distill git log -50 # 89% fewer tokens repowise saved # what distillation saved you, in tokens and dollars 如果倉庫沒有命令，本文不會自行編造步驟，而是建議先閱讀「up to −96% tokens to load context  ·  −89% file reads  ·  −70% tool calls」，確認系統依賴、預設埠與首次初始化。

日常使用取決於專案文件。README 的「Your agent stops guessing」段落提到：repowise exposes ten task-shaped MCP tools to Claude Code, Codex, Cursor, VS Code and anything else that speaks MCP. Most tools are built around data entities (one file, one symbol), which forces agents into long chains of sequential calls.。設定檔、環境變數、權限與資料目錄只在來源明確時才會記錄；沒有寫出的預設值，應在測試環境驗證並保留回滾副本。 同一部分也提到：Test intelligence. Ingest coverage, find untested hotspots, and run only the。