I built LogSentinel because I needed to analyze Nginx/Syslogs logs with AI, but strict data policies prevented me from sending raw logs to OpenAI/Claude.
How it works:
It tails log files in real-time.
Masks PII (IPs, emails, credit cards) using Regex before inference.
Sends the sanitized context to a local LLM (Ollama running Llama 3) to find anomalies.
Stores patterns in SQLite to avoid re-analyzing known errors (caching).
It's an MVP, written in Python. I'd love to hear your feedback on the architecture or how you handle local log analysis securely.
Great idea. For me, the architecture looks solid.
Hi HN, I'm Aibek, a sysadmin from Kazakhstan.
I built LogSentinel because I needed to analyze Nginx/Syslogs logs with AI, but strict data policies prevented me from sending raw logs to OpenAI/Claude.
How it works:
It's an MVP, written in Python. I'd love to hear your feedback on the architecture or how you handle local log analysis securely.