Approved for development
Trojan Eliminator Agent
A proposed Windows-first assistant. It would explain a finding already made by an established antivirus or EDR product, then wait while the owner approves any change. Read Trojan Eliminator Agent.
Catalog
Two catalogs
AI Defense Agents and Softwares are separate lists. Every name is a working name and is not trademark-cleared. Nothing in either list is released. Nothing here is a download.
AI Defense Agents
Open AI Defense Agents. These five are not Softwares.
Not released
Secret Sentry
Would look through repositories, logs, and process environments for leaked credentials. Read Secret Sentry.
Not released
Patch Pilot
Would inventory Windows and macOS applications and missing updates, then propose a patch plan for the owner to approve.
Not released
Exposure Check
Would check open ports and weak settings only on a domain or devices the owner is explicitly allowed to test.
Not released
Security Briefing Desk
Would summarize advisories that already name their sources, in language a non-expert can use.
Softwares
Open Softwares. These five are not AI Defense Agents.
Not released
Phish Inspector
Would explain a suspicious message, link, or attachment using reputation lookups and header analysis. It would not delete mail on its own.
Not released
Agent Firewall
Would block an AI agent from a risky tool, file path, or outbound domain unless a person approves that step.
Not released
Prompt-Injection Shield
Would screen a document or page before it is given to an AI agent and flag text that looks like an injected instruction.
Not released
Incident Notebook
Would help a small office keep an incident timeline, an evidence log, and a plain-language report.
Not released
Red-Team Range
Would exercise only the owner's own AI agents inside a sandbox and score the defenses. It is not for attacking anyone else's systems.
This description does not include an attack procedure. A later build, if one is ever approved, would be limited to systems the owner already controls.