🛡️ SENTINEL_WATCHER

SENTINEL_WATCHER is a high-performance .NET 8 Worker Service designed for privacy-first log monitoring. It bridges the gap between raw stack traces and actionable intelligence by analyzing local errors using Ollama LLMs and notifying you instantly via Telegram.

---

💡 The Concept

Sentinel operates on a simple Observe → Analyze → Alert pipeline:

1. Observe: Monitors local directories using FileSystemWatcher with minimal CPU overhead.
2. Filter: Identifies new entries marked as Error, Critical, Fatal or Exception.
3. Analyze: Sends the raw error context to a local Ollama instance. The AI interprets the stack trace and suggests a fix.
4. Alert: Pushes a notification report to your Telegram Bot.

---

✨ Features

• Privacy-First: Your logs never leave your local machine (No OpenAI/Anthropic keys required).
• Lightweight: Built on .NET 8 BackgroundService (Worker Service).
• Customizable: Works with any Ollama-supported model (llama3, qwen2.5, mistral).
• Real-time: Near-zero latency between log write and notification.

---

🚀 Getting Started

Prerequisites

• .NET 8.0 SDK
• Ollama (Running locally)
• Telegram Bot Token & Chat ID

Installation

1. Clone the repo: git clone https://github.com/youruser/sentinel_watcher.git
2. Pull your preferred model: ollama pull qwen2.5:0.5b
3. Configure appsettings.json (see below).
4. Run: dotnet run

Configuration

{
  "Ollama": {
    "Uri": "http://localhost:11434",
    "ModelName": "qwen2.5:0.5b"
  },
  "LoggerWorker": {
    "Path": "./logs",
    "Extension": "*.log"
  },
  "TelegramOptions": {
    "TelegramToken": "YOUR_TELEGRAM_TOKEN",
    "TelegramChatId": "YOUR_CHAT_ID"
  }
}

License

• MIT LICENSE