Hot Swap is a focused cabinet about the agent transfer protocol. Transfer an AI-agent core between robot bodies — each chassis moves differently — to solve spatial puzzles. It belongs to the AI & Robotics floor because it turns a subject from the AI Agents library into a short, repeatable decision loop. You can play without an account or installation, then return to the page to read the reasoning behind the mechanics.
How to play
Start with the default state and read the screen before making a change. The cabinet uses keyboard, touch controls, but the important control is the question you ask yourself before the next action: what is the system telling me, what am I trying to change, and what will this choice make harder? Read the state of the machine before you make the next move. The cabinet is built to make permissions, transfer, timing, or embodied action visible as a sequence of decisions rather than a single score.
The first run is for orientation. Notice which values move together, which action has a cost, and what the end screen remembers. On the second run, choose one deliberate priority. Do not try to improve every measure at once; a clear constraint makes the result easier to interpret. If the cabinet offers a pause, inspection, journal, or replay step, use it. Those are part of the lesson rather than friction around it.
What the simulation actually models
The model underneath is a constrained system: a capability, body, or agent can only do what its current state and permissions allow. The point is not to imitate a production stack. It is to let a toy system expose the relationship between an instruction, a capability, an environment, and a human decision.
The cabinet is not a forecast, certification, clinical assessment, engineering design, or field procedure. Its value is narrower and more useful: it gives you a safe place to feel a relationship. In Hot Swap, the relationship to watch is between the player’s logic puzzle choices and the state that follows. A run that goes badly can still teach you where the margin was, which signal you ignored, or which assumption was too broad.
A better second run
Before replaying Hot Swap, write one sentence about the previous result. “I spent the scarce action too early,” “I optimized the visible meter and starved the hidden one,” or “I treated a lead as proof” is enough. Then change one decision and leave the rest alone. This is the small experiment that turns a game over into an observation. It also makes the game a useful bridge into the longer guidebooks rather than a decorative button beside them.
How it was built
Like the other arcade cabinets, it is a self-contained browser application with its own state, controls, and local save behavior. The interface keeps the important variables on screen so the player can connect a choice to a consequence without a server account or a hidden progression system. The landing page stays separate from the playable bundle so the game can remain quick and focused while this page carries the explanation, links, and safety framing. No account is required, progress stays in the browser, and the cabinet does not ask the player to touch a real system or submit personal data.
Read the idea behind the cabinet
If you want the slower version, start with AI Agent Acceptance Criteria: Defining Done Before Delegation and AI Agent Access Reviews: Keeping Least Privilege Current . Read for the model, not a perfect score: the guidebooks name the evidence, constraints, maintenance habits, or human context that a short session has to compress. Then come back and try the cabinet with one of those ideas in mind.
The useful outcome is not that the game tells you what to buy, believe, diagnose, or do in an emergency. It is that the next real question becomes more precise. That is what a good learning cabinet should leave behind: curiosity with a boundary, a model you can explain, and a reason to look at the evidence again.