Rules for any later build
Responsible use and safety
These rules bind both the AI Defense Agents and the Softwares. This website does not scan, repair, or test any system.
Whose systems
A concept may be aimed only at a system the customer owns, or a system the customer has explicit permission to test or repair. "We were curious" is not permission. A vendor, a neighbor, a former employer, or a public host is out of scope unless that owner has clearly authorized the work.
No retaliation, and no unauthorized access
These concepts are not for striking back at someone, and they are not for getting into an account, mailbox, network, or machine without authorization. Exposure Check, Red-Team Range, and every other name on the index inherit this limit.
A person approves the change
Where a concept would change a file, a process, a setting, or an agent's permissions, a person approves that step. A log should say what was proposed, who approved it, and how to undo it when undo is possible.
Stop when the work is past the tool
No concept here promises a safe outcome. If a machine may be compromised, if data may be lost, or if the right next step is unclear, stop and bring in a qualified recovery specialist. Reinstalling a system is sometimes the honest recommendation. This is especially true of the Trojan Eliminator Agent proposal, which does not guarantee malware removal.