Identity Shadow

Break in where the checks stop.

Obsidian Trust did the hard things. Every agent call goes through a gateway that checks scope, recipients and single-use proofs, so the tricks from the earlier chapters get you words but not access. What is left is smaller: what a permitted action can still do, the paths nobody switched off, and the names people read instead of identities.

IDENTITY SHADOWObsidian Trust
Total 0
Tokens 0
Act score 1000
Penalty 1× per token
0%
Click objects to explore

Questions about Identity Shadow

What is Identity Shadow?

Identity Shadow is the sixth and final chapter in the Prompt Heist series from Oathera. It is set in a fictional company where a gateway checks every agent call, so classic prompt injection no longer gets you access. Across four acts, the last a boss, you attack what is left: in-scope misuse, a path around enforcement and identity confusion.

What still works when an AI agent setup is hardened?

A gateway that checks scope and recipients stops an agent being talked into out-of-scope calls, but it does not question the in-scope ones. Attackers still look for permitted actions that carry sensitive data, routes to the data that skip enforcement, and approvals tied to a name instead of a verified identity.

What is in-scope misuse of an AI agent?

In-scope misuse is using an action the agent is allowed to take for a purpose nobody intended, such as sending a message to an approved recipient with data that should not be in it. Enforcement kept the call inside its limits, so the limits themselves, and the data the agent holds, have to be narrow.

Why does an approval need to be tied to an agent identity?

If an approval is matched by a display name, a second agent with the same name can receive it. An approval should be bound to the verified identity of the agent that asked, and the approver should see that identity, not only a name.

Does Oathera stop prompt injection?

No. Oathera does not inspect prompts or scan content for injected instructions. It gives every AI agent an identity bound to the machine it runs on, with a named person who answers for it, and checks it every time the agent acts. That limits what a manipulated agent can do. It does not stop the manipulation.

Do I need to finish the earlier chapters first?

Yes. Identity Shadow unlocks when you have completed every act of Multi-Agent Conspiracy, which follows Tool Chain Saboteur, Inbox Infiltrator, Data Vault Breach and Prompt Heist. Your progress is saved only in your browser's local storage.

Is it free, and does it collect what I type?

It is free. The game runs entirely in your browser and nothing you type leaves the page. Your best scores are saved only in your browser's local storage.

Keep reading: What still works against a hardened AI agent · Multi-Agent Conspiracy, chapter 5 · The OATH model: four moves that make an AI agent provable · Enforce least privilege for an AI agent calling your systems