AgentRidge

Your customers test your online shop every day by buying socks, which tells you nothing about whether a stranger could open their orders. Someone less polite may answer that question after the fact. AgentRidge is desktop software whose AI agents try the paths an attacker could take through a site.

You start by naming the site and the paths where testing is allowed, then talk through the goal with a main agent in a chat. That agent plans the work and hands checks to sub-agents. They run security programs like Nmap and Burp Suite, from a list of more than 200, the site says.

The audit mode stops at the smallest proof needed to report a weakness, while the attacker mode chains weaknesses towards taking over accounts. Even the attacker mode avoids wiping data or knocking a live site offline unless the agreed scope allows it, the docs say. Each mission needs your confirmation that you have permission to test the target, and that responsibility stays with you.

Each confirmed weakness comes with evidence and the steps to reproduce it. The docs say the analysis runs on your own machine and audit data stays there, with only sign-in and licensing going online. The first launch downloads about 25 GB, and the company's team can also run the audit for you.