Security middleware for AI Agents. Intercepts shell commands before execution using a multi-layer pipeline: binary allowlist, regex patterns, deterministic intent coherence mapping, and LLM semantic check as last resort.

No description available

Έργο της ομάδας MindLab-AI Guardian για τον 6ο ΠΑΝΕΛΛΗΝΙΟ ΔΙΑΓΩΝΙΣΜΟ ΑΝΟΙΧΤΩΝ ΤΕΧΝΟΛΟΓΙΩΝ ΣΤΗΝ ΕΚΠΑΙΔΕΥΣΗ

Zero-trust access simulator with owner-only vault protection, dynamic policy enforcement, audit logging, and Labyrinth deception routing.

The total ai bot Guardian suite by SparkPit Labs

Web Scraper to extract article titles, publish dates, authors, and full texts from The Guardian News website. Built the program for McGill AI lab News Bias in Geopolitics research group