tg-alerts

Telegram Error Alerts

Adds operational error alerts to any project via a dedicated Telegram bot. Alerts go to a private channel, group, or forum topic — not to end users. Battle-tested pattern with deduplication, HTML formatting, and crash-proof error handling.

When to Use

• User asks to add error notifications, Telegram alerts, or operational monitoring
• Setting up a new deployment and want to know when things break
• NOT for: user-facing bot messages or interactive Telegram features

Setup Flow

This skill is interactive. Each phase involves asking the user questions or guiding them through Telegram actions.

digraph flow {
    rankdir=LR;
    "Assess\nProject" -> "Create\nBot" -> "Get\nChat ID" -> "Generate\nCode" -> "Integrate" -> "Test";
}

---

Phase 1: Assess Project

Detect from codebase or ASK:

1. Language — Python / Node.js / TypeScript / other
2. Framework — FastAPI, Django, Flask, Express, NestJS, Hono, etc.
3. Async or sync — determines implementation variant
4. Already has a Telegram bot? — alert bot MUST be separate from any main bot

---

Phase 2: Create Alert Bot

Guide the user through these Telegram steps:

1. Open Telegram, find @BotFather
2. Send /newbot
3. Name: something like "MyApp Alerts"
4. Username: something like myapp_alerts_bot
5. BotFather replies with a token — copy it
6. Save as ALERT_BOT_TOKEN in .env

ASK the user to confirm they have the token before proceeding.

---

Phase 3: Get Chat ID

ASK: "Where should alerts go?"

Target	Best for	Complexity
Private chat	Solo dev	Easiest
Group	Small team	Easy
Channel	Read-only broadcast	Medium
Forum topic	Organized by category	Medium

Guide based on their choice:

Option A: Private Chat

1. Open Telegram, find your new alert bot
2. Send it any message (e.g., "hello")
3. Run in terminal:

curl -s "https://api.telegram.org/bot<YOUR_TOKEN>/getUpdates" | python3 -m json.tool

4. Find "chat": {"id": 123456789} — that positive number is your ALERT_CHAT_ID

Option B: Group

1. Create a Telegram group (or use existing)
2. Add the alert bot to the group
3. Send any message in the group
4. Run the getUpdates curl above
5. Find chat.id — negative number like -1001234567890

Option C: Channel

1. Create a channel (or use existing)
2. Add the alert bot as administrator (needs "Post Messages" permission)
3. Post any message in the channel
4. Run getUpdates
5. Find "channel_post" -> "chat" -> "id" — negative, starts with -100

Option D: Forum Topic

1. Create a group -> Settings -> enable Topics
2. Add the bot to the group
3. Create a topic (e.g., "Errors")
4. Send a message inside that topic
5. Run getUpdates
6. Find TWO values:
   • chat.id -> ALERT_CHAT_ID
   • message_thread_id -> ALERT_THREAD_ID

Verify IDs

Always test before writing code:

curl -s -X POST "https://api.telegram.org/bot<TOKEN>/sendMessage" \
  -d chat_id=<CHAT_ID> -d "text=Test alert" -d parse_mode=HTML

For forum topics, add -d message_thread_id=<THREAD_ID>.

If the message appears in Telegram, proceed to Phase 4.

Troubleshooting empty getUpdates:

• Send another message, then run curl immediately (updates expire)
• For channels: bot must be admin, not just member
• For groups: send /setprivacy to @BotFather -> set to Disabled

---

Phase 4: Generate Alert Service

Use the appropriate reference implementation from this skill's directory:

Stack	Reference File
Python async (FastAPI, aiohttp)	references/alert_service_async.py
Python sync (Django, Flask)	references/alert_service_sync.py
Node.js / TypeScript	references/alert_service_node.ts

Read the reference file and adapt to the project:

• Match code style, naming conventions, import patterns
• Use the project's existing HTTP client if available
• Place where services/utils live in the project structure
• Add env vars to the project's config system

---

Phase 5: Framework Integration

Wire alert service into the project's error handling. The pattern:

1. Initialize on startup (only if env vars present)
2. Attach logging handler to root logger (captures all logger.error() / console.error())
3. Add unhandled exception handler specific to the framework
4. Send lifecycle alert on startup and shutdown

Framework	Unhandled Exception Hook	Startup Hook
FastAPI	@app.exception_handler(Exception)	lifespan context manager
Django	Custom middleware + 500 handler	AppConfig.ready()
Flask	@app.errorhandler(Exception)	App factory / before_first_request
Express	app.use((err, req, res, next) => ...)	After app.listen()
NestJS	@Catch() exception filter	onModuleInit()

---

Phase 6: Environment Variables

Add to .env and .env.example:

# Telegram operational alerts (optional)
# ALERT_BOT_TOKEN=123456:ABC-DEF...
# ALERT_CHAT_ID=-1001234567890
# ALERT_THREAD_ID=42

Ensure the config treats these as optional — alerting gracefully disables when not configured:

alerting_enabled = bool(ALERT_BOT_TOKEN and ALERT_CHAT_ID)

---

Phase 7: Test

1. Start the app with alert env vars set
2. Check for startup lifecycle alert in Telegram
3. Trigger an intentional error (e.g., hit a broken endpoint)
4. Verify error alert appears with traceback
5. Trigger same error again — verify dedup blocks the duplicate
6. Stop the app — verify shutdown alert

---

Alert Message Format

Consistent across all stacks — use Telegram HTML parse mode:

🔴 <b>Error</b>
<pre>Env: production
Error: ValueError
Context: [module.name] error message

Traceback (most recent call last):
  ...</pre>

🟡 <b>Warning</b>
<pre>Env: production
Some warning message</pre>

🟢
<pre>Application started
Env: production</pre>

---

Critical Design Rules

Non-negotiable for every implementation:

1. Alert failures NEVER crash the app — wrap all sending in try/except or try/catch
2. Log alert failures at DEBUG — ERROR would trigger the handler again -> infinite loop
3. Deduplicate by exception location — key: {ExcType}:{lineno}@{filename}. One alert per crash site
4. Truncate to 3500 chars — Telegram limit is 4096; stay well under
5. HTML-escape all dynamic content — tracebacks contain <>& breaking Telegram HTML
6. Separate bot — never reuse the main app bot for alerts
7. Include environment — every alert shows dev/staging/prod
8. Make alerting optional — app runs normally without alert env vars
9. Fire-and-forget — async: create_task(), sync: daemon thread. Never block requests
10. Cap dedup set at ~10k — clear when full to prevent unbounded memory growth

Common Mistakes

Mistake	Consequence	Fix
Same bot for app + alerts	Alerts fail when app bot is down	Separate bot token
No dedup	Hundreds of identical alerts	Dedup by exc type + location
await send_alert() in request path	Slow requests if Telegram lags	Fire-and-forget
Log send failure at ERROR	Infinite alert loop	Log at DEBUG
No truncation	Telegram silently drops message	Truncate to 3500 chars
Raw HTML in traceback	Broken formatting	html.escape() all content
Alert required for startup	App won't start without Telegram	Optional via alerting_enabled
getUpdates returns empty	Can't find chat ID	Send message first, curl immediately