Openclaw With Apple

Apple iCloud 全功能访问 + Apple Health 深度健康分析 + 双向待办同步

What to consider before installing: - Do NOT paste your Apple ID main password or one‑time 2FA codes into chat history. The SKILL.md explicitly encourages that — it's unsafe because chat logs may persist. Prefer local interactive login on a machine you control. - Prefer using an Apple app‑specific password whenever possible (the docs say calendar can work with that alone). If you only need calendar/CalDAV, avoid giving the main password. - Inspect scripts before running: review setup_tasks_cron.py, icloud_auth.py, tasks_tool.py and any code that writes cron entries, uploads files, or calls external endpoints. Only run on a personal, non‑shared machine. - The skill will cache session tokens under ~/.pyicloud/ — protect that directory (permissions) and delete/revoke sessions when you stop using the skill. Revoke app‑specific passwords on appleid.apple.com after testing. - The SKILL.md forces immediate command execution when the user mentions actions; consider disabling or modifying that behavior (require explicit confirmation) before using the skill broadly. - If you must use it: run the code locally (not on a cloud/shared server), create a separate Apple account for testing if possible, and audit the files the skill creates (tasks_latest.json, notes_latest.json) and scheduled cron jobs. Why I'm suspicious: the code matches the claimed capabilities but the runtime instructions ask for sensitive secrets via chat, enforce autonomous immediate execution, and the registry metadata fails to declare those credential requirements — this combination raises significant safety and coherence concerns. If you want, I can highlight the exact lines in scripts that persist sessions or add concrete steps to run the skill safely in a sandboxed/local environment.

Type: OpenClaw Skill Name: openclaw-with-apple Version: 1.0.0 The skill bundle provides a comprehensive suite for integrating Apple iCloud services, including Photos, Drive, Calendar, and Find My, as well as Apple Health analysis. It is classified as suspicious due to several high-risk behaviors: it explicitly instructs the AI agent to collect the user's Apple ID and Main Password (storing them in environment variables), establishes persistence on macOS via a LaunchAgent (setup_tasks_cron.py), and includes capabilities to track device locations and enable 'Lost Mode' (icloud_tool.py). Furthermore, the SKILL.md contains aggressive 'Iron Rules' that command the AI to execute tools immediately without user confirmation upon detecting specific keywords, which increases the risk of unintended actions via prompt injection. While these features are aligned with the stated purpose and documented in SECURITY.md, the combination of credential collection, persistence, and automated device control represents a significant security surface.