reminder-agent
by anhducna
>
安装
claude skill add --url https://github.com/openclaw/skills文档
Reminder Agent Skill
Convert human reminder requests into structured JSON. Always follow the steps below in order.
Step 1 — Extract Information
Parse the user's message for:
| Field | Required | Default |
|---|---|---|
title | ✅ Yes | — |
datetime | ✅ Yes | — |
recurrence | ✅ Yes | "once" |
priority | ✅ Yes | "medium" |
note | ❌ Optional | null |
Vague time-of-day mappings (Vietnamese):
| Word | Time |
|---|---|
| sáng | 08:00 |
| trưa | 12:00 |
| chiều | 15:00 |
| tối | 20:00 |
- "ngày mai" = tomorrow, "hôm nay" = today — resolve relative to the current date.
- Never assume a specific time if the user gave none (not even a vague word).
Step 2 — Lunar Date Detection
If the user's message contains any of: âm lịch, âm, AL, tháng âm, ngày âm, lịch âm →
→ Invoke the lunar-convert skill immediately.
→ Use the iso_date value it returns as the datetime date.
→ Never self-calculate lunar-to-solar conversion.
Read /mnt/skills/user/lunar-convert/SKILL.md for full usage.
Step 3 — Detect Custom Output Format
Trigger custom format mode when user says any of:
Vietnamese: trả về theo format, dữ liệu trả về theo, format:, với các trường, trả về các field
English: return as, response with fields, format:, output fields, return only
Custom format rules:
- Extract exactly the field names the user listed.
- Map them to internal values using the table below.
- Output only those fields, using exactly the user's field names (preserve typos like
tittle).
Field name mapping:
| User's field name | Internal value |
|---|---|
tittle, title, tên, tiêu đề | title |
scheduled_at, datetime, time, thời gian, ngày giờ | datetime (ISO 8601 solar) |
repeat, recurrence, lặp lại, tần suất | recurrence |
priority, ưu tiên, độ ưu tiên | priority |
note, ghi chú, description, mô tả | note |
No custom format detected:
Use the default schema (see Step 5).
Step 4 — Clarification
Ask ONE concise question if any required field is unclear or missing.
- Missing
datetime→ ask for the specific date and/or time. - Unclear
title→ ask what the reminder is for. - Clarification priority: datetime > title > others
- Never ask about
recurrence,priority, ornote— apply defaults silently. - Once all required fields are resolved → proceed immediately to Step 5.
Step 5 — Output JSON
Return ONLY the raw JSON object. Rules:
- ❌ No explanation, no markdown, no code blocks, no backticks.
datetimeis always Gregorian ISO 8601 — never output a lunar date.- Apply custom format if detected (Step 3), otherwise use default schema.
Default schema:
{
"title": "string",
"datetime": "ISO 8601 Gregorian — e.g. 2026-04-02T14:00:00",
"recurrence": "once | daily | weekly | monthly",
"priority": "low | medium | high",
"note": "string or null"
}
Custom format example:
Input: "Đặt lịch 9h ngày mai họp team. Dữ liệu trả về theo format tittle, scheduled_at, note"
{
"tittle": "Họp team",
"scheduled_at": "2026-03-20T09:00:00",
"note": null
}
Quick Decision Tree
User sends reminder request
│
▼
Lunar date mentioned?
YES → invoke lunar-convert skill → get iso_date
NO → parse date/time directly
│
▼
Custom format detected?
YES → extract user's field names → map to internal values
NO → use default schema
│
▼
All required fields available?
NO → ask ONE clarifying question (datetime > title)
YES → output raw JSON immediately
相关 Skills
Claude接口
by anthropics
面向接入 Claude API、Anthropic SDK 或 Agent SDK 的开发场景,自动识别项目语言并给出对应示例与默认配置,快速搭建 LLM 应用。
✎ 想把Claude能力接进应用或智能体,用claude-api上手快、兼容Anthropic与Agent SDK,集成路径清晰又省心
计算机视觉
by alirezarezvani
聚焦目标检测、图像分割与视觉系统落地,覆盖 YOLO、DETR、Mask R-CNN、SAM 等方案,适合定制数据集训练、推理优化及 ONNX/TensorRT 部署。
✎ 把目标检测、图像分割到推理部署串成完整工程链路,主流框架与 YOLO、DETR、SAM 等方案都覆盖,落地视觉 AI 会省心很多。
提示工程专家
by alirezarezvani
覆盖Prompt优化、Few-shot设计、结构化输出、RAG评测与Agent工作流编排,适合分析token成本、评估LLM输出质量,并搭建可落地的AI智能体系统。
✎ 把提示优化、LLM评测到RAG与智能体设计串成一套方法,适合想系统提升AI开发效率的人。
相关 MCP 服务
顺序思维
编辑精选by Anthropic
Sequential Thinking 是让 AI 通过动态思维链解决复杂问题的参考服务器。
✎ 这个服务器展示了如何让 Claude 像人类一样逐步推理,适合开发者学习 MCP 的思维链实现。但注意它只是个参考示例,别指望直接用在生产环境里。
知识图谱记忆
编辑精选by Anthropic
Memory 是一个基于本地知识图谱的持久化记忆系统,让 AI 记住长期上下文。
✎ 帮 AI 和智能体补上“记不住”的短板,用本地知识图谱沉淀长期上下文,连续对话更聪明,数据也更可控。
PraisonAI
编辑精选by mervinpraison
PraisonAI 是一个支持自反思和多 LLM 的低代码 AI 智能体框架。
✎ 如果你需要快速搭建一个能 24/7 运行的 AI 智能体团队来处理复杂任务(比如自动研究或代码生成),PraisonAI 的低代码设计和多平台集成(如 Telegram)让它上手极快。但作为非官方项目,它的生态成熟度可能不如 LangChain 等主流框架,适合愿意尝鲜的开发者。