
What happened
Developer zer0 released AlienPentest for Kali Linux, which runs five phases — recon, scan, exploit, post-exploit and report — after you enter a target IP and press LAUNCH. It needs a purchased license key and a Claude API key.
Why it matters
Tasks that once required running nmap, nikto and gobuster by hand and reading each output can now be chained automatically, so pentest work may need far less manual tool operation and specialist time.
WHO IT HITSSecurity consultants and in-house pentesters who currently run reconnaissance and vulnerability tools by hand could hand that chain to the AI, while learners can watch its reasoning on practice targets like Metasploitable2. Anthropic API usage is billed directly to the user per scan.
Summaries like this, every morning.
The pitch rests on how penetration testing is normally done: a tester runs tools like nmap, nikto and gobuster one at a time and interprets each output before choosing the next move, which takes deep expertise and a great deal of time. AlienPentest replaces that loop with AI judgment, moving from open ports and running services to known vulnerabilities and then to whether entry is actually possible, and finishing with a report that includes recommended countermeasures.
The distribution model is split in two. The license itself is bought from alienpentest-license.fly.dev and comes in STR, PRO and ENT tiers that differ on monthly sessions, whether exploitation is allowed, and PDF export; the AI side runs on a Claude API key obtained from Anthropic's console, with usage billed by Anthropic and consumption per scan varying with target size. A stealth toggle that inserts 0.5–3 second gaps between tool runs and caps concurrency is offered for real engagements where detection matters, and an interrupted scan can be resumed from a button on the next launch.
For practice, the tool is paired with Metasploitable2, an intentionally vulnerable VM, and lists the flaws it is meant to find automatically — a backdoor in vsftpd 2.3.4 that yields a root shell, a backdoor command in UnrealIRCd 3.2.8.1, an OS command route in Samba 3.0.20, password-free root access to MySQL 5.0.51a, and code execution through Apache PHP 5.2.4 URL arguments. The author states that every operation described was carried out only in environments they own or manage, and the tool is presented as being for certified professionals holding written permission, given that unauthorized testing is a criminal offense under laws including Japan's Unauthorized Access Prohibition Act and the U.S. Computer Fraud and Abuse Act.
Pick your industry and the AI tools you use, and get news related to your work every day.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
Meta rolled out specialized versions of its existing models, trained using Muse Wonder, to Muse and Meta AI us…

Mastercard joined a $15m Series A round for firmly, an AI shopping agent firm, becoming a payments partner as…

Salesforce says its three-person Rapid Innovation Squads delivered a Sales Signals Slack app in days that has…

Uber Eats is seeing little traffic from AI agents, according to Nation’s Restaurant News

ServiceNow is expanding partnerships to roll out agentic AI and large-scale AI workflows across enterprise cus…

Uber detailed Uber Secure AI, a defense-in-depth framework with three integrated layers — provisioning-time va…
