3sk1nt4n/cybersentinel-ai — reverse-engineered prompt

Build me a local cybersecurity assistant that I can run with Docker and open in the browser. I want a clean dashboard with a chat style interface where I can ask it to scan a target I own, and it should actually run real security tools in an isolated sandbox instead of just suggesting commands. It should support things like port scanning, web checks, vuln lookups, SQL injection testing, DNS and subdomain recon, OWASP style testing, and threat intel lookups, then explain the results in plain English as they stream in.

Make it work fully offline by default with a local model, but let me optionally plug in other AI providers and switch between them without losing the conversation. I also want the app to use a knowledge base for MITRE, NIST, CIS, and compliance guidance, show relationships in a graph view, and include SIEM style log analysis with sample security events. Please add sensible guardrails, cancel support for running scans, easy startup from the included scripts or Docker Compose, and keep it safe for authorized testing only.