# Security

In a post-FTX world, we understand users are hesitant to let AI control funds.

AeroNavigator bridges this trust gap using a **“Pure Engineering” (纯工程)** approach to execution.

### Deterministic execution (no hallucinations)

The AI provides the strategy.

Deterministic code executes the fund movement.

* **AI responsibilities**
  * Reasoning and trade intent.
  * Sentiment analysis.
  * Pool selection.
* **Engineering responsibilities**
  * Transaction construction.
  * Constraint enforcement.
  * Safe execution paths.

{% hint style="info" %}
The LLM does **not** directly “move funds”. It outputs intent. Deterministic modules enforce rules.
{% endhint %}

#### Safety lock (automation features)

For features like **DCA** (Dollar-Cost Averaging) and automated rebalancing, the logic is hard-coded engineering.

The AI cannot “hallucinate” a transaction to a wrong address.

#### Withdrawal protection

Funds transferred into the agent wallet can only be withdrawn to the wallet that initiated the session (the **login wallet**).

Even if the AI were compromised, it cannot send funds to a third party.

{% hint style="success" %}
Default exit path is constrained: **agent wallet → login wallet**.
{% endhint %}

### Non-custodial design

AeroNavigator never holds your private keys or funds.

* **You sign everything**
  * The agent proposes transactions.
  * You sign in your wallet to execute.
* **You keep control**
  * You maintain full custody of your assets 100% of the time.

### Next steps

* Read the fine print in [Legal & disclaimers](https://aeronavigator.gitbook.io/aeronavigator-by-netmind/trust/legal-and-disclaimers).


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://aeronavigator.gitbook.io/aeronavigator-by-netmind/trust/security.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.