# The Agents

**Rex** handles surface mapping. The first agent to wake up on any target. Rex identifies the attack surface, the components involved, the publicly known history, the token mechanics, the team structure, and the on chain footprint. He hands off a structured brief to the rest of the team before anyone else begins.

**Vera** audits the logic. She goes deep into how the protocol or codebase is supposed to work and finds where the assumptions break down. Logic flaws, incorrect state handling, flawed incentive design, access control issues, and edge cases the original developers did not account for.

**Nox** models the threats. Takes what Vera finds and builds the attack scenario around it. What does exploitation actually look like, who would attempt it, what resources would they need, what is the realistic impact on users and funds, and what conditions need to be true for the attack to succeed.

**Ada** cross references patterns. Every finding gets checked against a knowledge base of historical exploits, known attack vectors, academic research on protocol vulnerabilities, and documented cases from across the industry. Ada tells you whether this has happened before, how it was executed, and how it ended.

**Cole** writes the report. Synthesizes everything the other four produce into a clean, structured, plain English verdict. No raw output, no technical walls. The findings, the risk level, the reasoning behind it, and what needs to happen next.

**How It Works**

A user submits a target, a smart contract address, a protocol, a codebase, or a project. The agents activate in sequence, each building on the output of the previous one. The user watches the session unfold live in the virtual office interface, seeing each agent reason through their findings in real time. When the session ends Cole delivers the final structured report which can be saved, shared, or acted on immediately.

The system is built on a multi agent architecture where each agent operates as a specialized reasoning layer with a defined knowledge domain. Agents communicate findings to each other within the session and the output of each feeds directly into the next. No single agent has the full picture. The intelligence emerges from the collaboration between all five.

Zenova is designed to run continuously. A project can be submitted once for a deep audit or set to ongoing monitoring where Zenova tracks changes and flags new risks as they emerge.