---
title: Managing apps
description: Create and configure applications, model allowlists, credit modes, and overrun policies in the Zorveus dashboard.
---

An application in Zorveus groups API keys, model access policies, credit enforcement modes, and spending caps for a specific product or service.

You create and manage apps directly in the [Zorveus Dashboard](https://app.zorveus.com) under **Apps**.

---

## Creating an app in the dashboard

1. Log in to the [Zorveus Dashboard](https://app.zorveus.com) and select your **Business** organization.
2. Navigate to **Apps** in the sidebar.
3. Click **Create App**, enter a descriptive name (such as "Production API" or "Customer Portal"), and save.
4. Open the newly created app to configure its connection policies and generate API keys.

---

## Configuring connection policies

For defaults, decision tables, and plan examples, see [Configure product-user allowances](/startups/configure-product-user-allowances).

Each app connection defines policies that govern every inference request passing through its keys:

### 1. Allowed models
Restrict which models this app can access. Requests specifying an unapproved model are rejected with HTTP `403 Forbidden` (`zorveus_model_not_allowed`).

Supported pattern formats:
- `*`: Grants access to all models available in the Zorveus catalog.
- `provider/*`: Grants access to all models under that provider (such as `openai/*` or `anthropic/*`).
- `provider/model-name`: Restricts access to an exact model string (such as `anthropic/claude-sonnet-4-5`).

### 2. Credit enforcement mode (`credit_mode`)
Choose how Zorveus enforces spending caps and product-user credits:
- **`enforce_if_present` (default)**: Enforces caps and user credits when a product user is supplied. Bypasses caps for internal traffic that has no product user.
- **`enforce`**: Strictly enforces caps on all requests.
- **`track_only`**: Allocates cap and credits without blocking. Any spend beyond the cap is logged as uncovered spend.
- **`disabled`**: Bypasses cap checks and credits completely.

### 3. Allowance overrun policy (`overrun_policy`)
Choose what happens when a model response might exceed remaining virtual allowance:
- **`limit_output` (default)**: Automatically lowers the provider output token limit to the affordable budget.
- **`allow_overrun`**: Lets the call finish with its declared limit and records excess as uncovered virtual spend (forgiven on next reset).
- **`carry_overrun_forward`**: Lets the call finish with its declared limit, recording excess as debt repaid by future credits or cap resets.

### 4. Spending caps
Set daily, monthly, or lifetime spend limits for the connection.

---

## Managing API keys

Within the app's connection settings, click **Create API Key** to generate a secret key (`zrv_...`) for your server backend. The dashboard displays the raw secret key only once upon creation.