New in the Loop with AI Pentesting

To ensure safety at Vulnetic, there is a major feature we have pushed into the loop, which I will share here. I am hoping this becomes a biweekly log to update users on not only what new features, but why we pushed them.

One of the most requested features was the ability to switch between user input and autonomous mode to ensure safety in production environments. To enable this, we have created REPL. This allows the user to give approval before each command, ensuring the human is in the loop. When the user is satisfied with the methodology of our autonomous agent, they can quickly press the gear icon and watch Sable hack autonomously.

With a quick switchback to agent mode, Sable continues the deductive penetration testing path. After executing the command you approve, the REPL field will be populated by the next command, waiting for user approval to execute.

REPL also allows not only command injection but prompting as well. A \# symbol followed by a prompt can direct the task in the direction you want. We see this as a great way for apprehensive users to get involved with our product, increasing your productivity while still being highly restricted.

If you are interested in trying out Sable, contact us at contact@vulnetic.ai or visit our website at Vulnetic.ai — The AI Pentester