Add testing skill for Telegram bot integration tests

.agents/skills/testing-telegram-bot/SKILL.md

@@ -0,0 +1,100 @@
+---
+name: testing-telegram-bot
+description: Test the refactored Telegram bot end-to-end. Use when verifying bot module changes, bug fixes, or new features.
+---
+
+# Testing the Telegram Bot
+
+## Overview
+
+The bot is a modular Python Telegram bot under `bot/` with 12 modules. Testing can be done at two levels:
+1. **Offline integration tests** (no BOT_TOKEN needed) — covers DB, web server, proxy fetch
+2. **Live Telegram tests** (requires BOT_TOKEN) — covers conversation flows, file streaming
+
+## Devin Secrets Needed
+
+- `BOT_TOKEN` — Telegram bot token (only needed for live Telegram interaction tests)
+- `SUPER_ADMIN_ID` — Telegram user ID for admin (can use dummy `999999` for offline tests)
+
+## Offline Testing (No Token Required)
+
+### Setup
+```bash
+cd /home/ubuntu/repos/Turbo-Window-RDP-DEV-
+pip install -r requirements.txt
+```
+
+### Key Pattern: Monkey-patch DB_NAME
+The `Database` class uses `DB_NAME` from `bot.config` (no constructor arg). To test with a temp file:
+```python
+import bot.database as db_mod
+db_mod.DB_NAME = "/tmp/test.db"  # or tempfile.mktemp(suffix=".db")
+from bot.database import Database
+db = Database()
+await db.init_db()
+```
+
+### Key Pattern: Set env vars BEFORE importing bot modules
+```python
+import os
+os.environ["BOT_TOKEN"] = "1234567890:ABCdefGHIjklMNOpqrsTUVwxyz"
+os.environ["SUPER_ADMIN_ID"] = "999999"
+# NOW import bot modules
+```
+`bot/config.py` calls `sys.exit(1)` if BOT_TOKEN is empty, so it must be set before any bot import.
+
+### What Can Be Tested Offline
+
+1. **Database layer** — All CRUD operations, schema creation, migrations
+2. **Web server endpoints** — Build a minimal `aiohttp` app with handlers from `bot.server`
+3. **Proxy fetching** — `bot.handlers.proxy.fetch_all_proxies()` hits real public APIs
+4. **Module imports** — Verify no ImportError across all 12 modules
+5. **Conversation states** — Verify count matches expected (currently 19, 0-indexed 0-18)
+
+### Web Server Testing Pattern
+```python
+from aiohttp import web
+from bot.server import proxy_api_handler, health_handler
+
+app = web.Application()
+app["db"] = db  # your test Database instance
+app.router.add_get("/api/proxies", proxy_api_handler)
+app.router.add_get("/health", health_handler)
+
+runner = web.AppRunner(app)
+await runner.setup()
+site = web.TCPSite(runner, "127.0.0.1", 19876)
+await site.start()
+# Test with aiohttp.ClientSession against localhost:19876
+```
+
+### Common Schema Gotchas
+- `file_links` table: `file_type` is NOT NULL — always include it in test INSERTs
+- `user_logs` table: column is `details` (not `detail`)
+- `discount_codes`: `expires_at` should be set to future date for valid codes
+
+## Live Testing (Requires BOT_TOKEN)
+
+With a real token, you can test:
+- Bot startup via `python run.py`
+- Conversation flows (send /start, interact with menus)
+- File streaming via `/stream/{f_uid}` endpoint
+- Broadcast messaging
+- VIP payment flow
+
+The bot entry point is `bot.main.main()` (not `build_application`).
+
+## Key Bug Fixes to Verify on Changes
+
+1. **Referral balance**: `start.py` should show `referrer['balance']` (already includes bonus from DB)
+2. **Discount timing**: `validate_discount()` must NOT increment `used_count`; only `use_discount()` should
+3. **Proxy API limit**: `server.py` wraps `int(limit)` in try/except, defaults to 50
+4. **Session cleanup**: `server.py` registers `on_cleanup` handler for ClientSession
+5. **File download count**: `get_file()` is read-only; `increment_file_downloads()` is separate
+
+## Running the Test Script
+```bash
+cd /home/ubuntu/repos/Turbo-Window-RDP-DEV-
+python test_integration.py
+```
+Expected: 44/44 assertions pass. The test script sets dummy env vars automatically.