Detects attacks in application logs using DBSCAN clustering, PCA, and local LLM analysis.
Cluster HTTP and system logs using PCA and DBSCAN, then analyze outliers with local Ollama LLMs.
Webhawk (Catch 3.0) is an open-source security tool that identifies attack traces in web server and system logs without pre-written rules. It applies unsupervised machine learning to cluster log entries, flagging anomalous activity for automated LLM review. SOC analysts and compliance teams can use it to surface hidden threats across endpoints.
The engine converts raw logs into numerical features via PCA and applies DBSCAN density clustering to isolate outliers. Anomalous logs pass to a local LLM running via Ollama to generate incident descriptions and cross-reference known CVEs. Detections stream directly to a web interface or integrate with existing SIEM environments through a REST API.
It suits security analysts, incident responders, and GRC teams looking to catch zero-day attack patterns across Apache, Nginx, or process logs. It is not suitable for non-technical teams or environments lacking dedicated compute for self-hosted AI models.
Groups routine log entries by feature density and flags outliers as potential attacks without signature rules.
Routes detected anomalies to Ollama models to generate context-aware incident summaries and advice.
Queries the NVD database to correlate detected log anomalies with published security vulnerabilities.
Provides a Python agent to read local access log files and ship them to the central detection engine.
Webhawk is free and open-source under the MIT license. The primary cost comes from providing the server hardware required to host the Docker stack and run local LLMs.
877 open remote AI-training roles, updated November 16, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
These reviewed AI tools do the same job instantly — from our scored directory.
It includes built-in parsing regex patterns for Apache access logs, Nginx access logs, HTTP logs, and OS process details.
No, default configurations route log analysis to a self-hosted Ollama instance running local models like Granite4.
See Zekai first in Google