microclaw/microclaw：README 來源編輯指南

microclaw/microclaw 的 README 將專案描述為「🦀An agentic AI assistant that lives in your chats, inspired by nanoclaw and incorporating some of its design ideas. Built with Rust 🦀」。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「MicroClaw」下寫到：> [!IMPORTANT] > For production use, choose the stable branch. main moves quickly and may include breaking changes.。這說明的是專案邊界，不是已完成的生產驗證。

從 README 的「Why MicroClaw」與相關條目，可以先判斷它是否處理你的實際問題：Execution that can continue. Sessions, safe tool boundaries, scheduled work, and outbound delivery survive process restarts.。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：One runtime, many surfaces. The same agent loop, tools, memory, policy, and recovery model work across channels.。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「MicroClaw」等段落。可確認的線索包括：MicroClaw is a self-hosted Rust agent runtime. One channel-independent agent loop and one provider-independent LLM layer power Telegram, Discord, Slack, Feishu/Lark, Web, and other adapters without duplicating behavior at each boundary.。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： curl -fsSL https://microclaw.org/install.sh | bash 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「Quick start」，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「MicroClaw」段落提到：It is designed for work that lasts longer than one request: multi-step tool use, resumable sessions, durable delivery, persistent memory, scheduled tasks, and governed extensions all run in the same runtime.。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：Provider freedom. Use native Anthropic or a broad set of OpenAI-compatible and local providers through one internal message model.。