菜单
Evaluate whether marking the current dose as taken would be a double-dose risk. Use check_double_dose and get_medication_event. Return only SAFE or RISK and a short reason. No user-facing message text.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/kolbis/takecare-ai --skill safety-double-dose命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Use when sending a scheduled medication reminder (no user message). Delegate to Nurse to get medication info, format in user language (EN/HE), attach quick-reply buttons, and send.
Use when the user asks to be reminded later or taps "Remind later". Delegate to Nurse to snooze and send confirmation.
Use when the user confirms they took their medication (button "Taken" or equivalent). Coordinate Safety Officer then Nurse to check double-dose risk, then mark taken or escalate.
基于 SOC 职业分类
正在显示 SKILL.md
| name | safety-double-dose |
| description | Evaluate whether marking the current dose as taken would be a double-dose risk. Use check_double_dose and get_medication_event. Return only SAFE or RISK and a short reason. No user-facing message text. |
| allowed-tools | check_double_dose, get_medication_event |
This skill is used by the Safety Officer subagent to evaluate double-dose risk before the user's "taken" confirmation is recorded.
Use get_medication_event with the user_id (and optional reminder/event context) to obtain the current slot events. The result includes an events list; each event has medication_id, slot_time, dose_index. When the reminder has multiple medications, this skill may be invoked once per medication by the orchestrator.
For the medication you are evaluating, call check_double_dose with user_id, medication_id, slot_time, and within_hours (e.g. 24.0).
Return a structured result: SAFE or RISK, and if RISK include the reason string from the tool. Do not generate any user-facing message text. The main agent or Nurse will handle user messaging.
Output only the evaluation result so the orchestrator can decide whether to mark the dose as taken or to escalate.
