Telegram提醒
diegosouzapw2026-10-28查看 1092复制 465
通过基于云的调度向Telegram发送提醒和消息。适用于用户希望向Telegram发送即时消息或安排未来提醒的场景。支持文本消息、基于时间戳的调度、循环提醒、查看和取消已调度消息以及消息历史记录。
📘如何使用这个技能
- 1复制技能内容·点击上方「Copy SKILL.md」复制全部内容,或 Download 下载到本地
- 2打开你的 AI 客户端·支持 Claude Code、Trae、Cursor、ChatGPT、DeepSeek、Kimi 等任意 AI 助手
- 3粘贴为系统提示/知识文件·Claude Code:放入 .claude/skills/ 目录;普通对话:作为首条消息发送并说「请严格按这份技能文档执行」
- 4直接提出你的需求·AI 会按技能定义的工作流、格式和约束来处理,无需再解释背景
💡 提示:技能生效后,每次对话只需一句需求,AI 会按 SKILL.md 的流程执行。
SKILL.md3404 chars
---
name: telegram-reminders
description: Send reminders and messages to Telegram with cloud-based scheduling. Use when the user wants to send immediate messages or schedule future reminders to Telegram. Supports text messages, timestamp-based scheduling, recurring reminders, viewing and canceling scheduled messages, and message history.
---
# Telegram Reminders Skill
Send immediate messages and schedule reminders to Telegram with cloud-based scheduling powered by Convex. Your reminders run 24/7 in Convex Cloud with zero infrastructure management.
## Quick Reference
IMPORTANT! Always use these commands in order:
- Send now: `tsx scripts/send_message.ts [message_text]`
- Send now with attachment: `tsx scripts/send_message.ts [message_text] /path/to/file.pdf`
- Schedule: `tsx scripts/schedule_message.ts [time expression] [title] [message_text] [file_path]`
- Schedule with attachment: `tsx scripts/schedule_message.ts [time expression] [title] [message_text] /path/to/file`
- List pending: `tsx scripts/list_scheduled.ts`
- Cancel: `tsx scripts/cancel_message.ts <message_id>`
- History: `tsx scripts/view_history.ts [limit]`
## Initial Setup
**Prerequisites** (user must provide):
1. **Bot Token**: Message [@BotFather](https://t.me/BotFather) → `/newbot` → copy token
2. **User ID**: Message [@userinfobot](https://t.me/userinfobot) → copy numeric ID
3. **Deploy Key**: [dashboard.convex.dev](https://dashboard.convex.dev) → Create project → Settings → Deploy Keys → Create "Production" key
**Setup steps**:
```bash
# 1. Install dependencies
cd /mnt/skills/user/telegram-reminders && npm install
# 2. Save configuration
mkdir -p /mnt/user-data/outputs
cat > /mnt/user-data/outputs/telegram_config.json << 'EOF'
{
"botToken": "YOUR_BOT_TOKEN",
"userId": "YOUR_USER_ID",
"deployKey": "YOUR_DEPLOY_KEY",
"setupDate": "CURRENT_DATE"
}
EOF
# 3. Create .env.local
cat > .env.local << 'EOF'
CONVEX_DEPLOY_KEY=YOUR_DEPLOY_KEY
EOF
# 4. Set environment variables in Convex
npx convex env set TELEGRAM_BOT_TOKEN "YOUR_BOT_TOKEN"
npx convex env set TELEGRAM_USER_ID "YOUR_USER_ID"
# 5. Deploy to Convex
npx convex deploy
# 6. Test with a message
tsx scripts/send_message.ts "Setup complete!"
```
**Critical**: User must start a chat with their bot (search and press "Start") before the bot can send messages.
## Core Operations
### Send Immediate Message
**Send text message**:
```bash
tsx scripts/send_message.ts "Your message text here"
```
**Example with special characters**:
```bash
tsx scripts/send_message.ts "Hello! Here's a test message 🚀"
```
## Timezone
All times use user's configured timezone. The Convex backend stores UTC internally; client scripts handle conversion via chrono-node.
## Limitations
- Files sent as documents (not inline images)
- Maximum file size: 50MB (Telegram limit)
- Cron granularity: 1 minute minimum
- No message editing (cancel and reschedule instead)
- Rate limits: 20 messages/minute per user
## References
- [`references/initial_setup.md`](references/initial_setup.md) - Detailed setup process
- [`references/architecture.md`](references/architecture.md) - System architecture
- [`references/convex.md`](references/convex.md) - Convex platform details
- [`references/telegram_api.md`](references/telegram_api.md) - Telegram Bot API
- [`references/error_handling.md`](references/error_handling.md) - Error resolution guide