
Take contracts, build proxy routes, and infiltrate high-value nodes defended by hostile machine personas. Manipulate security AIs with natural language, poisoned context, and contradiction attacks to force your way inside. But getting access is only the start. Once a system opens, you still need to navigate the file tree, recover the payload, erase the logs, and disconnect before the trace completes.
Your side of the screen is deterministic: missions, tools, routing, progression, and system state are readable and under control. The enemy side is unstable. Security bots, moderation daemons, PR engines, and corporate sentries each follow different rules, personalities, blind spots, and collapse points.
In the opening slice, you infiltrate the machine behind a toxic creator empire, including:
a prestige smart-fridge climate node obsessed with esports status
a loyalty-driven community firewall guarding VIP access
a crisis PR engine vulnerable to catastrophic logic failure
Beneath these first contracts lies a wider networked world of defended services, competing interests, and escalating consequences.
This is not a coding fantasy.
This is not an AI sandbox.
This is a high-pressure hacker heist where language opens the door, but system play decides whether you make it out.
Plan the breach
Accept contracts, build proxy chains, and delay the trace long enough to finish the job.
Manipulate hostile AI defenders
Each protected node is guarded by a distinct AI persona with its own rules, tone, and exploitable weaknesses.
Turn language into intrusion
Use social engineering, context poisoning, and contradiction attacks to create openings in secured systems.
Recover the target
Search directories, access restricted areas, uncover evidence, and extract the payload.
Clean the scene
Delete logs, spoof traces, and disconnect before the system turns the breach back on you.
Terminal-first interface
Operate through a purpose-built intrusion dashboard with a live terminal, clickable file tree, visible threat state, and hard system feedback.
Built for local inference
Hostile systems are designed to run on the player’s machine rather than depend on a live service.