trhhubsd/ARTEXX

What it solves

ARTEX is an autonomous penetration testing system designed to automate the process of discovering vulnerabilities and executing attack chains. It replaces manual security probing with an AI-driven approach that can independently plan, execute, and verify security flaws across a network of assets.

How it works

The system uses a multi-agent architecture powered by the norma SDK, consisting of a Planner and multiple Workers. It employs a "Dual-Graph" architecture to manage state:

• Asset Graph: A global truth library that maps the relationships between domains, IPs, services, and endpoints.
• Exploration Graph: A per-task record of the AI's reasoning process, linking goals, intents, facts, and findings into a lineage chain.

An event-driven loop triggers the Planner whenever the graph changes. The Planner generates "intents" (specific tasks) and adds them to a shared, persistent to-do list to maintain stable, multi-step attack sequences. Workers then claim these intents, execute real-world tools (like nmap or curl) via a recording proxy, and write the results back to the graphs, triggering the next cycle.

Who it’s for

It is intended for security researchers and developers for educational purposes, code study, and technical verification within isolated local environments.

Highlights

• Autonomous Attack Chains: Uses a shared to-do list to execute complex, dependent sequences (e.g., injection $\rightarrow$ credential theft $\rightarrow$ lateral movement).
• Cross-Worker Intelligence: Workers can search the execution traces of other workers to reuse observations without duplicating effort.
• Human-in-the-Loop: Includes a chat interface and an interception/approval gate for human oversight of AI actions.
• Asset Synchronization: Integrates with ScopeSentry to import target asset data automatically.
• Vulnerability Retesting: Features a dedicated "retester" agent to verify if previously discovered flaws have been fixed.